chemical physics of discharges - Argonne National Laboratory

)i
1
reesonably complete and accurate synthesis to describe the macroscopic character-
istics. This approach of course implies that one has a detailed understanding of
the processes involves and know tile rates and cross sections for the processes.
While one can use this approach succesfully up to a point, the extreme complexity
4 of discharge phenomena forces one also to formulate a macroscopic description
, and work toward a juncture with the microscopic point of view.
1
Chemical Physics of Discharges
\.lade L. Fite
University of Pittsburgk
Introduction
_I___
The gas discharge as a chemical tool is of interest for the environment
which it provides, an environment which in many respects is very close to and in
others very far awa:: from thermal and chemical equilibrium. Since achieving +.his
situation in tne gas discharge is a result of the ph:rsical processes occurrir,g to
sustain the electrical characteristics of the discharge it is appropriate to
consider sone of the 5asic physical aspects of gas discharges before examining
the chemical consequences.

2
I1 Electron Motion and Behavior
Be it supposed that a gas is enclosed in a container and that a free electron
has been produced--perhaps by a cosmic ray, or by the radiation fro4 the experimenter's '
watch dial, or from that little piece of uranium glass in a graded seal, or by I
cold cathode emission by the very strong electric fields around a sharp point on
an electrode in the discharge tube.
I
If a dc electric field is imposed, the electron will respond according to \
Newton's law,
-&€
"= nf
and begin to accelerate in free fall. This acceleration will continue until it
has a collision with a gas molecule. If the energy of the electron at the time
of collision is very low, only elastic scattering will be admitted. At higher
energies large amounts of energy can be lost by the electron in exciting the molecule
to high-lying states of internal energy and at higher energies yet, ionization can
occur. The latter process is, of course, essential to achieve the electron multi-
plication and convert the gas with a single electron in it into a gaseous medium
of high electrical conductivity.
Since each type of collision has separate consequences for the discharge ve
consider them in turn. The collisions are of course statistical and we must
inco_rporate statistics with particle mechanics in their treatment.
A. Elastic Collisions: Transfer of Energy
In elastic collisions of electrons with heavy gas molecules, two effects are
of importance. First there is a redistribution of directions of travel of the
electrons and second there is a very slight loss of energy (and therefore speed)
as a very small mount of momentum and energy are transferred to the molecule by
the electron.
At low energies, i.e., a few ev and less, electron scattering tends to be
isotropic in angle in the center of mass coordinates. This implies that if we
examine many electrons immediately after they have had a collision the average
vector velocity will be zero, but of course the average scalar speed will not be
zero. The probability per unit time that an electron will have a collision
(i.e. the collision frequency) is given by
wtiere n is the number density of molecules, c is the electron's speed and Q(c)
is the total cross section for elastic collisions. The cross section generally
diminishes with increasing speed, although resonances in scattering impart
considerable structure in the Punctional form of the cross section in the case
of most gases.
We suppose that an electric field, E, is imposed in the z direction, and
(1)

3
inquire about the component of velocity in the z direction, averaged over all
electrons, , which is identical to . Since each electron responds to the
field in the same way, the average acceleration due to the field isthat of
each electron separately. Further, because the average velocity is 'zero following
a collision, the loss of average velocity due to collisions is to good approximation
just the product of the average collision frequency and the average velocity.
Thus we can write that for the average velocity in the z direction
When the steady state is achieved the left side vanishes, which implies that the
pickup of directed velocity from the field is just balanced by the loss of directed
velocity due to the randomization of direction of motion by the collisions. Under
steady state collisions , then,
where p is designated the mobility.
Equation (4) can be re-written in terms of Eq. (2) as
In this expression the second term gives all the information about the interaction
between the electron and the particular gas molecule and the third term contains
the parameters available to the experimenter, i.e., the electron field and the
gas number density. Since the number density is proportional to the gas pressure,
the experimental quantity of major relevance is the ratio E/p.
Randomization of directions of velocity says nothing whatever about average
speeds of the particles. For this we turn to other considerations. Conservation
of momentum requires that the energy loss from an electron of mass m and kinetic
energy W in collision with a heavy particle of Mass M is given by

4
where 0 is the angle of deflection of the electron in the collision. If the
scnttering is isotropic (as it close1.v approximates at low energies) then the
energy loss in a collision averaged over angle is ,
where for convenience we let 2m/M = h.
We would expect a steady state ultimately to be achieved between the electron
energy picked up from the electric field between collisions and the energy transmitted
from the electron tc the heavy particles through the collisions. We proceed to
evaluate these separately. For simplicity we assume that the collision frequency
is independent of speed of the electrons.
Since the r'--tric field acts only in the z direction, all the increase in
kinetic energy of an electron will occur through increasing the z component of its
velocity which is given by
where voz is the value of vz immediately following the collision and t is the time
elapsed since the last collision and a = (q/m)E. The energy picked up will be
If we now average over all angles of direction of motion immediately following the
last collision, '-0 so that
If we now average over all collision times, Eq. (10) becomes
I
(9) I ,
1

4
I
I
5
To find (t2)av, we assume that collisions are random events and the collision
frequency is independent of speed. We can then write that the probability of a
collision occurring in the interval t to t + dt following the precdding collision is
. .
for which the average value of t2 is 2@ where< = 112 is the mean collision time.
Eq. (11) then becomes
a&#
t2.
= ma-t
If we also average over all speeds of electrons and neglect some of the finer pohts
of statistics, Eq. (13) becomes
where A is the mean free path and c is the mean speed of the electrons.
This quantity, in the steady state, will equal the right hand side of Eq. (7)
averaged over all electrons. If we let d.ls 1 -2 c
2
Thus we would expect that the mean speed would be given by
-
and the mean energy of the electrons in the steady state by
(13)

6
The remarkable aspect of this result comes on the insertion of numbers in
Eq. (17). If for example, one has electrons moving through a gas with M 30 AMU,
at pressures giving X ?r Imm (i.e. p 6 1 torr) then the mean energy in ev is around
25 times the field strength in volts/cm. With as little as a few tenths of v/cm
fields, mean electron energies are several ev. We can thus understmd the appearance
of high electron "tempreatures" in gas discharges.
It is beyond the scope of this review to go further and question the distribution ,
of electrcn energies. It is sufficient to indicate that if one takes assumptions
similar to those nade in this simplified argument and utilizes them in the Boltzmann
ea-uation, one can obtain an approximate solution for the distribution of speeds. The t
result is not the I.laxwell-Boltzmann distribution, but rather that known as the
Druyvesteyn distribution whose dependence on speed is given by
This distribution function is similar in shape to the hxwell-Boltzmann distribution
for a given mean speed, except that the most prob2ble speed is slightly higher and the
high energy tail is diminished in the Druyvesteyn distribution. Nonetheless, the
distributions are sufficiently similar that one can, to good approximation, think of
the electrons as having a temperature in the Maxwellian sense which is much higher than
the neutral gas temperature; i.e. typically 30,000'K vs 300'K. It is this dichotomy of
temperatures which is perhaps the most striking of the non-equilibrium aspects of a gas
discharge.
This general effect of collisions randomizing the direction of motion of electrons
which have picked up energy from the field between collisions has another important
manifestation. It is responsible for the operation of microwave electrodeless
discharges.
E(t) = Eo cos ut, then Newton's law for the electron becomes, in the absence of
collisions,
which solves to give
If we consider an electron moving in an ac field which is of the form
The rate of energy pickup from the field i.e. the power, P is given by
tEv
4
<
\

\
vnich, it is noted can be negative as well as positive. If we consider the total
power pickup over a comFlete cycle, the sine-cosine product integrates to zero,
indicatinfi that there is no net transferral of energy from the electric field to
the electron. This is of course a result of the fact that the applied field and
the electron velocity are 90O out of pnase.
7
Furthermore, we can note from (20) that the maximum electron velocity will be
a_Cg/nu and the maximum kinetic energy will ue
It is noted that if one has a field oscillating at say 109 cps and a maximum field
strength of say 300 volts/cm, equation (22) indicates that the maximum kinetic energy
of the freely oscillating electron will be only about 2 ev. "his is insufficient
enera to ionize gases, and yet breakdown will occur for this type of field.
It is collisions of the type which lead to the Druyvestryn distribution which
are responsible for ionization. An electron will be accelerated by the field during
a portion of its cycle and then be deflected in a collision. The energy parallel
to the electric field is thus converted to energy perpendicular to the field.
The field then proceeds
to give the electron additional energy in the direction of the field. Energy is
pumped from kinetic energy of motion parallel to the field to kinetic energy in all
directions.
If we write Newton's law for an electron, adding a collision term, we are left
with Eq. (3), except now, E is time-dependent. Writing E = Eo cos ut
1
I - p = I - fiz - (24)

Again if this power gain is equated to the loss of energy of electrons in
elastic collisions with the gas molecules, as would occur in the steady state, then
qEo
where a f -. m
E
8
From (26) one would expect the electron "temDerature" for fixed Eo to diminish
with increasing frequency, but for pressures of the order of one torr and gas masses
of around 30 AMU the electron temperature will remain near the dc value for frequencies
usually employed for gas discharge work.
B. Elastic Collisions, and Diffusion
-
Since in either dc or ac discharges, the electrons will have high temperatures,
one can expect that all the manifestations of a kinetic temperature will be present.
Particularly important among these is diffusion through elastic collisions. If only
electrons are present in a neutral gas they will tend to diffuse as would any other
gaseous component with a current density
where ne is the number density of the electrons and De is the diffusion coefficient of
the electrons through the gas.
will contain a mobility term as well so that
I
(25)
If a field is also superposed, the current density \
,
where the minus sign in the second term indicates that because of the negative charge 4
on the electrons, their motion will try to be in the direction opposite to that Of
the applied field. pe is taken to be a positive number. Whether the electron motion
is diffusion-dominated or field dominated depends on whether the first term
l
is larger
'1

I
7
than or smaller than the second.
In a gas discharge, the electrons will have produced ions and :these also will
tend to diffuse through the neutral gas as well as respond to any electric fields.
The current density of ions will be given therefore by a similar equation
--3
s, = -9+ On, 4 “+F+
9
From Eq. (4) the mobility of either type of particle is & Ifl[m(z>)
and from kinetic theory, the diffusion coefficient is given by D e. KT/t-m).
Since both these parameters have the mass appearing in the denominator the electron
will diffuse and respond to an electric field much more rapidly than will the heavy ions.
The case of major interest for discharge physics is that where the number density
of ions is very nearly equal to that of the electrons. Clearly because the electron
mass is very light these will try to diffuse awa:y from the ions. However, as soon as
they do this, an electric field is set up between the electrons and ions so that the
electrons are held back by the ions and the ions are dragged along by the electrons.
We would thus expect that the current fluxes Se and S+ would be equal.
If we set Se = S, 5 S, and ne = n+ t n, to indicate an electrically neutral
plasma, Eqs. (28) and (29) can be combined to eliminate the electric field with the
result that
where
Since
and
or
and b z &
re ,gel
we can write
(29)

LU \
If the electron temperature is equal to the ion temperature, as in the case of
late in an afterglow, the ambipolar diffusion coefficient is just twice the value
of D+. In the active discharge, however, as we have seen Te will be large, perhaps
of the order of 30,000°K. On the other hand the ion temperature will deviate little
from the neutral gas temperature. The reason is that the ion mass id compwable
to the mass of the neutral molecules; thus by arguments similar to those leading
to Eq. (7) the ion will in a single collision be able very effectively to give to
the neutral gas the energy it picks up from the field between collisions. The
temperature of the ions will therefore remain very close to the neutral gas temperature.
The ratio Te/T+ will be of the order of 100 typically in an active discharge
and rapid diffusion of the plasma through the neutra1,gas and to the walls of the
container will result.
Summary of the Remainder of the Paper
C. Electron production and loss mechanisms are discussed and the plasma
balance equation is formulated.
D. Excitation to radiating and metaetable states is eumaarized and some of
their consequences for the operation of a discharge are presented.
E. Wall phenomena.
111. Macroscopic phenomena.
A.
Plasma polarization and the Debye length are discussed.
-B. Characteristics of certain types of discharges are reviewed; Glow Discharges,
Arc Discharges and rf discharges.
IV. Afterglows.
briefly reviewed.
The effect8 of suddenly reducing the electron temperature are

i
I
i
I
I
I
)
)
i
L
,
I
I . INT1:ODtiCTIOfi
1l
THE PRODIXTIOX OF 1\TOi.!S A::D SI:lTLL PJ$I)ICP.LS L< GLOW OISCIIARChS
*
Frederick Kaufman
Department of Chemistry, University of Pittsburgh
The size and topical variety of this symposiun clearly show tliat electrical
discharges are finding increasing application in many areas of cliemistry ranging
from tile iproduction of simple atoniic species such as 11, 0, or i from their diatomic
molecules to the sviitiiesis or specific ueconposition of complex ory.anic or inorgsiiic
ComPoUndS. It is unfortuiiately true that our understanding of tile chcmistrv of dis-
cllarge processes is still in a rudimeoLary state, tiint the field is more an art than
a science, and thus represents one of the last f'rontiers of chemistry.
There is jiood reason for this unsatisfactory state of affairs. Glow disciiar-
Res are complex phenomena in which gases at sul,-atmosplit!ric pressure are undergoing
excitation and ionization by electron impact and so Give rise to hiziily uncquili-
brated steadv-state conditions where the effective temperature of free electrons is
typically tens of thousands OK, that of electronically or vibrationally excited
States may be thousands of OK, whereas tiie transl.ationa1 and rotational temperature
will only be teris to hundreds of '1; above ambient. It sliould be clear, of course,
that apart from the processes occurring at tiie electrodes, energy from the electric
field is coupled to the xas almost eiitirelv through the kinetic energy of free
electrons wiiicli, due to their small mass, acquire eiiergv more rapidlv froni the field
and lose it more slowly in elastic collisions (the mean fractional energy loss per
elastic collision equals 2 ci/N in the siiiiplest clasical model where in and X are the
masses of the electron and of the molecule). In tllis manner, electrons become suf-
ficiently cnercetic to ionize some of the neutral species ana thereby balance their
continuous loss by diffusion, attachment, and recombination. As the ionization
potentials of most neutral gases are'in the 10 LO 20 ev range (230 to 460 kcal/mole),
an appreciable fraction of the electrons has enough energy to produce electronic
excitation (responsible for the emitted glow) and dissociation.
In the following sections, the mechanism of dc and ac glow discharges will be
briefly described, with emphasis on high frequency electrodeless discharges (f = lo6
to 1O1O sec-') and on the simple geometry often encountered in rapidly pumped steady-
state flow systems at pressures near 1 torr. After a brief discussion of tile rates
and energy dependence of specific collision and diffusion processes, available ex-
perimental data will be brought to bear on the problem of H2, S2, and 02 dissociation
and on the chemistry of some more complicated systems.
Although there are several fine monographs available on electron impact plie-
nomena and, discharge physics'-', thev contain relatively little information on active
high frequency discharges which is pertinent to the problem of dissociation and chem-
ical reaction. The electron physics of microwave discharges is discussed in some
review articles. 5*6
XI. BASIC PtiYSICAL PROCESSES
IT. 1. General Mechanism and Frequency Dependence.
Glow discharges are typically observed in the pressure range of about 0.1 to
10 torr. .At much lower pressures, the electron mean free path is too long for gas
collisions to be important, electrons pick up large amounts of energy from the dc or
slow ac field and bomhard the anode or the tube wall which may then fluoresce. At
* This work was supported by tiie National Science ?oundation
/I

L
12
much higher pressures, the mean free path is very short, the breakdown field strength
is very high, and when it is exceeded, local, highly ionized, but narrow pathways are
created for the conduction of current, i.e. spark filaments are formed. The normal
dc glow discharge in a long cylindrical tube is characterized by many axially distinct
but radially fairly uniform regions of quite different optical and electrical proper-
ties such as the Aston Dark Space, Cathode Glow, Cathode Dark Space, Negative Glow,
Faraday Dark Space, Positive Column, Anode Glow, and Anode Dark Space, in this order,
between cathode and anode. The reason for this complexity is that quite different
processes OCCUK in the different regions as is also shown by a very,non-uniform
voltage rise between cathode and anode. Host of the potential difference is taken
up in the 'cathode fall' which comprises the first four regions enumerated above, is
dependent on the cathode material as well as on the nature of the gas and on the
natural variable E/N (V cm*/molecules) where E is the field strength (V/cm) and N the
total neutral density (molecules/cm3). The latter is a measure of the electron
energy under conditions of steady-state drift. Equating energy loss and gain per
2m cV 1
electron per second, - - - eEw, where E = - mT2 is the electron energy, Q the mean
M e 2
free path for electron-neutral collisions, e the electronic char e, and w ty
eEQM1$' =- E eM1,'
electron drift velocity which equals 3 % one obtains c = (6m)'f2 N 211uL(3m)n
where Q was replaced by lql/ 211u2N)from simple kinetic theoj.. The 'cathode fall'
and 'anode fall' regions a so have large gradients of electron and ion concentrations
and a local imbalance of electrical charge. The positive column is simpler in nature
(although striated positive columns are still poorly understood), has a small and
constant axial voltage drop, and only a small imbalance of charge carriers, because,
although electrons initially diffuse to the tube wall faster than ions, the resul-
tant radial field prevents further charge separation and forces electrons and ions to
diffuse equally fast. This process is called ambipolar diffusion and is further
discussed below. The voltage drop along the positive column is also independent of
the total current over a fairly wide range, and since the current, 1, is carried
mostly by the electrons whose drift velocity is about 100 times larger than that of
the ions, i = n ew = - - n which shows that the electron concentration, ne,
3 mC e'
increases linearly with increasing current because E and J are constant. In its
normal range of electron (and ion) concentrations of lo8 to 10" ~ m - ~ the , positive
column of a dc glow discharge is diffusion-controlled and serves as the electrical
connection between the cathode and anode regions.
In high frequency electrodeless discharges the complications of the cathode
and anode regions are absent, the entire plasma is approximately neutral and
diffusion-controlled, and the discharge often resembles the positive column of an
equivalent dc discharge. Yet, there are differences in its fundamental mechanism,
especially at microwave frequencies. Free electrons oscillating in an alternating
field can not derive power from the field on the average, because their motion is
90' out of phase with the field. They therefore acquire energy only because c01lisions
with neutral molecules change their phase relationship with the field, while
at the same time representing a small fractional loss of the energy gained. Under
the assumption that the ac frequency, f, is greater than the elastic collision frequency,
v the maximum displacement, x, of an electron due to the high frequency
e'
2eE
field is given by x = where w = 2llf. In an active microwave discharge E is
typically 30 V/cm and x is therefore less than cm when f = 2.5 x lo9 sec-l (as
in the widely used Raytheon Microthem Generator). The corresponding maximum electron
energy acquired during the cycle is eEx, about 0.02 ev, i.e. the electrons
slowly accummulate the energy necessary to undergo inelastic, ionizing CollIsiOnS
and to sustain the discharge. When elastic collisions are approximately accounted
for, it is convenient to define an effective field strength, Ee = Eo ("2 + ,f
where E
.2 )1/2
is the rms value of the applied field strength. The power gained from the
9

13
field per electron is and per collision ttrereforc
11. 2. iliffusion of Ciiargeci Species.
2 e2L 2 ,
in u
e
rnGT *
In the discliarges of interest iiere, tile concentration of charged species is
greater ellan and a large fractional separation of elrctrong alid positive ions
becomes impossible, because it would set up a very large opposiug field. The currents
of electrons and ions reaching the wall must tiicn be equal, i = - i) Vn - nu E'
i+ = - D+Vn + nu E' where n = n
+
IJ the diffusion coefficient, LI the mobility, and E' the field due to tile (small) space
charge. The subscripts e and + refer to clectror. and positive ion. Equating i and
i, and eliminating C' one obtains
r
Dew+
I = -
+ U+we
?+ + Ue
= n+ is the electron density, 011 the densitv gradient,
Vn = 1) Vn
which serves as the definition of the anbipolar diffusion coefficient, Da. SubstitukTe
ting P - for i)
e c
and the equivalaut expression for 1)+ otic obtains
v+k
which approximately equals - ('1
c e + T ) or D+(1
+
T
+ ")
T+
because the electron mobility
pee' is much larger than the ionic mobility P+. In active glow discharges T /T+ is
typically 20 to 100, whereas in the afterglow the electrons thermalize rapidly ana
Te/T+ = 1.
Thus, D 2 20 to 100 D+ in the active discharge, and 2 D+ in the afterglow.
The disappearance of charged species bv ambipolar diffusion in the absence of
6n
a source term is described by the diffusion equation = uavZn where n is a function
of r, 8, z, and t. The well-known solution of this equation for the case of an infinite
m
cylinder is n(r,t) = Z
-kit
AiJo(% I) e where al is the i th root of Jo, the Bessel
i=1 0
a.
2 Da
function of zero order and k = ($) . The diffusion length, A, therefore
i
Da =
r 0
equals . The first few zeroes of Jo are a1 = 2.405, a2 = 5.520. a3 = 8.654,
'i
a,, = 11.792 which shows that, as diffusion proceeds, tile time decay will be increasingly
governed by kl = (-) Da, the first (lowest) diffusion mode, because the next three
r0
higher modes are damped out more rapidly by factors of 5.3, 12.9, and 24. After a
short transient, the diffusion-controlled electron decay or the diffusion controlled
loss under steady-state conditions with a spatially well distributed source term can
therefore be closely approximated by a first-order rate constant, k = 5.78 Da/ro2.
This analysis applies when there are only positive ions and electrons present.
h'hen negative ions are also present, their principal effect is to accelerate the
ambipolar diffusion of the electrons, (Da)e, which now becomes approximately
T
Te
(Da)e = (1 + A) D+ (1 + -) + X - 1) which for active discharges (Te/T+ >> 1)
*+ T+
can be further approximated by (1 + 2 A) Te/T+, where
ratio of negative ions and electrons.
= n-/ne. the concentration
It can be seen that for A >> 1 electrons

.
14 \
will be lost much more rapidly be diffusion to the wall than in the absence of
negative ions. t
11. 3.
Electron-Ion and Ion-Ion Recombination.
Although several radiative, two-body, and three-body recombination mechanisms
exist, only the fastest ones will be mentioned here. These are the dissociative recombination
of electrons and positive molecular ions as exemplified by
NO+ + e + N + 0 (where the products are likely to be electronically’ excited), similar
ion-ion reactions such as 1; + I- + I2 + +
I or 31, NO + NO2- + neutral products, and ,
+ -
three-body ion-ion recombinations such as NO + NO2 + M + neutral products.
All of these processes have very large rate constants, due to the long range
coulombic attraction between reactants, if reasonable paths are available for the
dissipation of the large exothermic reaction energy (* 10 ev) such as dissociation 1
and electronic excitation. The first of these three processes has been studied in
greatest detail, especially by Biondi and coworker^^'^ who found a value of I
2.9 5 0.3 x lo-’
3 -1 -1
cm molecule sec for the rate constant of Np+ + e * N + N, for 1
example. The generally observed range of 1 to 5 x lo-’ (6 x 1013 to 3 x 1014 lit 1
mole-l sec-l) means that at an electron and ion concentration 10l1 which is near)
the upper limit of charged particle densities encountered in glow discharges, the
effective first order rate constant for electron removal under steady-state conditions,
1
will be 1 to 5 x lo4 sec-l.
Two-body ion-ion recombinations have rate constants in the same general range
although few have been studied in detail, none with precise analysis of reactants and
products. Some three-body ion-ion recombinations have recently been studied by Mahan
and coworkersg’ lo who found effective termolecular rate constants in the range
4 x 1U-26 to 3 x
approximate dependence, and at a total pressure of 1 torr, such processes
would have effective first-order rate constants of ion removal in the 10 to 100 sec-’
range, too slow to be of importance.
11. 5.
+
cm6 molecule‘2 sec” for NO + NOp- + M near 300’K. With an
Electron Attachment and Detachment.
Only the fastest of the many possible processes need to be discussed here.
Radiative attachment and photodetachment as well as three-body attachment processes
are unlikely to be of importance. Dissociative attachment reactions such as
e + 02 + 0- + 0 have rate constants1’ which rise from zero at an electron energy
threshold (4 to 9 ev for the formation of 0- from 02, NO, or CO) to a maximum of
to cm3 molecule-l sec-l for electrons with 6 to 10 ev. For average electron
energies of 2 to 3 ev in an active discharge, the effective rate constant must therefore
be lowered about 10 fold to a range of to Moreover, several aSSOci
ative detachment reactions such as 0- + 0 + 02 + e, 0- + N + NO + e, 0- + 82 -+ 840
have recently been found12 to be very rapid (k = 1 to 5 x cm3 molecule’’ Sec-’)
under thermal conditions near 3OO0K. This further reduces the likelihood that negati
Ions are important species in rapidly pumped steady-state glow discharges of diatomic
gases. In this regard, active discharges probably differ markedly from their Cor-
responding afterglows in which the electrons are rapidly cooled to ambient temperatux
and will then readily attach to form 02-, NO-, Cog-, Cob-, and other negative ions
even though the electron affinities are quite small.
I
1
>
!

.
11. 5. CharRe-Transfer and Ion-Molecule Reactions
Both positive and negative ion-molecule reactions have recently been studied
by a variety of experimental methods, and consistent values for many rate constants
have become available. Because of the strong ion-dipole or ion-induced dipole interaction,
these reactions usually have little or no activation energy if they are exothermic,
and often have rate constants near lo-' cm3 molecule-' sect', in accord with
the simple theory based on the polarizability of the neutral reactant. Some excep-
tions such as 02+ + Ng +
NO+ + NO which is at least lo6 times slower are probably
due the large energy requirements for bond rearrangement which in the corresponding
neutral four-center reaction gives rise to a very large activation energv. Other unusually
slow reactions such as 0' + h2 -c NO+ + N (k 'L 2 x lo-") are less easily
rationalized, particularly since k rises sharply when the reactant 'L2 is vibrationally
excited.
From the magnitude of to lo-' for many of the exothermic reactions it is
clear that the effective first-order rate constant for the transformation of an ionic
species by reaction with a major neutral constituent (P = 1 torr = 2 x molecules
at the higher temperature of the discharge) is 2 x lo6 to 2 x lo7 sec-l, i.e.
such reactions will go to completion in a small fraction of the residence time in even
the most rapidly pumped flow systems.
Minor neutral constituents such as atomic
species at 0.1 to 1 male % will transform ionic species witti rate constants of lo3 to
lo5 sec-l, still much faster than the rate of traversal through most discharges whose
average flow velocities are in the range 102 to lo4 cn sec-' and whose lengths are 1
to IO cm.
11. 5. Electron Impact Ionization.
As shown in section 11. 2 above, the electron loss term by ambipolar diffusion
can be approximated by a first-order rate constant, k = 5.78 Da/ro2, which for an
cm, makes
active discharge with Te/T+ "i, 50, U+ % 100 cm2/sec, and ro = 0.5
k 2, 1 x lo5 sec-l. Although it is conceivable that chemi-ionization will occur in
which two electronically excited molecules with 5 to 10 ev energy react to produce
ionization, such processes will normally be of very minor importance. Even with a very
large chemiionization rate constant of lo-'' cm3 molecule-' sec-l, a concentration of
5 x 1014 ~ m - ~ 2 , to 3 mole %, of such excited molecules would be required to balance
the diffusional loss. It thus seems likely that electron impact ionization is the
major source term for charged species in the discharge. Although this requires more
than the ionization potential of the atom or molecule, i.e. electron energies in excess
of about 15 ev, the large electron velocity and its very high average temperature,
Te, can easily provide the required magnitude of the rate constant.
The ionization cross section of most atoms and simple molecules rises sharply
from zero at the ionization potential and comes to a broad maximum of about 1 to 5 x
cm*/molecule at electron energies of 70 to 120 ev. It normally reaches a value
of 1 x about 2 to 4 ev above its ionization potential, i.e. at electron energies
of 12 to 17 ev. For average electron energies of 2 to 3 ev in active glow discharges,
this lends to total effective ionization rate constants of IO-" to loe1' cm3 molecule-'
sec-l if a Maxwell distribution is assumed. A somewhat lower range would be obtained
for a Druyvesteyn distribution, but since the ratio E/? of required to average energy
is never very large the error due to the assumption of a Maxwell distribution should
be fairly small.
An important, though infrequently applicable ionization mechanism is the direct
ionization'by collision with sufficiently energetic, metastable neutral species. This
process, called Penning ionization, can occur only if the excitation energy of the
metastable exceeds the ionization potential of the other reactant. For He, Ne, and Ar
the excitation energy of the lowest triplet state is 19.80, 16.62, and 11.55 ev re-
' spectively, and since the rate constants for Penning reactions are often gas kinetic,
I

state is bound,but the molecule is formed on the repulsive part of its potential energy 1
cume at a point above its dissociation energy, and will therefore dissociate on its first'
' 111. APPLICATION TO THE DISSOCIATION OF SOME DIATOMIC MOLECULES.
111. 1. Summary of Some Atom Production and Loss Processes
(1

17
In the followin:: sections much evidence will be cited for catalytic effects
in the production of atomic species such as 11, N, or 0 in glow discharges of t!ieir
diatomic gases. The existence of such effects suggests large ctlanzes in the atom
production or loss terms upon small variations in qas composition. Tile-principal
processes are therefore summarized in this section, and approximate ranges given for
their rates. A cylindrical discharge tube of 1 cm diameter, average electron concentration
of 10l1 ~ m - ~ and , electron energy of 2 to 3 ev are assumed corresponding
to known conditions in microwave discharges at input power levels of1 about 10 to 500
watt.
Production terms: (a) Electron impact dissociation via excited states de-
pends on the existence of a dissociating or predissociating state at moderate excita-
tion energy and is therefore quite variable. When the dissociation energy is rela-
tively small, as in €1 and (,,and when states are available at 8 to 10 ev excitation
threshold, an atom production rate of 10 to 100 torr/sec can be calculated. For N2
whose dissociation energy is larger, the rate will be at least 10 times lower. It
should be clear that this process must occur in active discharges, since electron
impact ionization, which requires appreciably larger electron energies, is the princi-
pal source term for electron production. Therefore, since there are enough electrons
present with 15 to 20 ev energy to balance the rapid ambipolar diffusion (and recom-
bination) loss, there must be appreciably more with 8 to 12 ev for excitation-
dissociation.
(b) Electron-Ion recombination at the wall (following ambipolar diffusion)
or in the:gas phase will also produce atomic species in reactions such as
e + 112' -+ 2 H, e + N2+ -t N + N, etc. -It is clear that: the upper limit to this atom
production term is given by the total rate of ionization except for a possible factor
of two from the stoichiometry of the dissociation. But as this surface recombination
may also lead to the molecular product by the dissipation of energy to the surface,
the dissociation rate due to this process is likely to be lower than the corresponding
ionization or equivalent ambipolar diffusion term, i.e. < 0.5 to 5 Torr/scc. The gas-
phase dissociative recombination will produce atoms at a rate of 0.1 to 0.3 Torr/sec
at an electron density of emq3 and much more slowly at lower densities.
(c) Ion-molecule reactions of primary ions may in exceptional cases be sufficiently
exothermic to produce atomic species and ions of lower ionization potential
such as in H2+ + H2 * l!3+ + 11, but the equivalent reactions for 02 and fi2 are endothermic.
When such a process is possible, its rate is again limited by the total rate
of ionization, but should closely approach it if the neutral molecule is a major species
as in the above ii2 reaction.
(d) Neutral-neutral dissociation reactions may also occur, but little can be
said about them in general. As the kinetic temperature of all but the electrons is
fairly low (300 to 800°K, mostly near 6UO°C), two excited, metastable molecules would
have to react to transfer their excitation to a predissociating state. Few such reactions
are known and their rate constants are likely to have upper limits in the
to cm' molecule-l sec-lrange. If that were so, metastable mole fractions
as high as 0.1% would only produce atoms at ratios of 0.02 to 0.2 Torr/sec. Such
states would also have to be optically metastable, i.e. have radiative lifetimes longer
than 0.01 sec, and be resistant to collisional quenching, i.e. have deactivation probability
per collision lower tlian
Loss Terms: (a) iiomogeneous sas-phase atom recombinations are three-body
processes vith rate constants near 10- cm6 molecule" sec-' so that even for atom
mole fractions of 5X, such loss rates are near 0.01 torr/sec, negligibly small.
(b) Surface recombination is a well-known process, usually kinetically first
order, and characterized by a rate constant, k = E , for a cylindrical tube of diameter d.

H.ere Y
18
is the recombination coefficient, the fraction of surface collisions leading
to recombination, and E is the average molecular velocity of the atomic species. y
which is often as low as outside the discharge for a well-cleaned or suitable
poisoned surface, is likely to be or larger in the active discharge. For y = 10-3,
atom loss rates of 2 to 10 torrlsec can be calculated, and for larger y, these rates
would be larger, but not proportionately so, because large radial concentration gradients
would be produced and diffusion would become rate-controlling. Then, an effective
first order rate constant, k = D/A2 - 5.8 D/r2, would be applicable, similar to ambipolar
diffusion (Section 11.2). but with the smaller, molecular diffpsion coefficient.
For an atom mole fraction of 52, the upper limit to the diffusion91 atom recombination
rate will be 100 to 300 torrlsec depending on the atomic species.
(c) Ion-molecule reactions can remove atomic species if the ion is polyatomic
such as in N + Ng+ * N2 + N2+ or the equivalent oxygen reaction. If the polyatomic ion
is a major ionic species, and the reaction is a fast one, its rate at 5% atom mole
fraction will be near 5 torrlsec. It will then $e limited by the rate of regenTration
of the polyatomic ion. Reactions of the type 02 + N * NO+ + 0 or Ng+ + 0 * NO + N
are known to be fast, but as they replace one atomic species by another they need not
be considered here.
(d) Neutral, bimolecular atom-molecule reactions such as N + 02 * NO + 0 or
0 + H2 + OH + H often have appreciable activation energies. Moreover, they, too, only
shuffle atomic species and can not be considered loss terms. In the few applicable
cases where this does not hold as in 0 + N20*2 NO or N + N20 * N2 + NO, the reactions
are known to be negligibly slow.
111. 2. Hydrogen Discharges
The discussion of the electron collisional aspects of H2, N2, and 02 discharges
is greatly aided by the excellent papers by Phelps and coworkers in which elastic and
inelastic collision cross sections are obtained by numerical solution of the Boltzmann
transport equation and comparison with all available experimental data on electron
transport coefficients. For H2 and D214, the vibrational excitation cross section
becomes appreciable
cm2) at about 1 ev and peaks near 5 ev at 8 x cm2.
For dissociation, a threshold at 8.85 ev and peak of 4.5 x cm2 at 16 to 17 ev
was found to be consistent with the data. A recent measurement by Corriganl’ of the
dissociation cross section of H2 is in good agreement with the above except for a
larger peak of about 9 x crn2 at 16.5 ev. For average electron energies, E of
3.0 and 2.0 ev, dissociation rate constants, kd, of 11 and 2 x 10-10cm3 molecules
kl 1
sec-lcan be calculated by summation of Qvf(s) from E = o to m, using energy increments
of 2 or 3 ev, where Q is the appropriate cross section, v the electron velocity, and
is the Maxwell distribution function. Such k ‘s correspond to H-atom production rates
d
of 200 or 40 torrlsec. An effective ionieation rate constant, k , was similarly
found to be 7 x
( E ~ = 3ev) and 6 x ( E - ~ 2 ev), corresponding to ionization
rates of 7 and 0.6 torrlsec, respectively. The latter should be approximately equal
to the ambipolar diffusion loss, 5.8/rO2 D+Te ne/T which equals 9 or 6 torrlsec if
g’
D+.= 700 cm2/sec at 1 torr pressure. This crude calculation suggests that E - 3 ev is
a better choice than 2.0 ev. It neglects the serious deviation from a Maxwelfl distribution
which is to be expected.
More realistic estimates of kd and ki were obtained by Dr. Phelps by simul-
taneously adjusting E/N, ck. and using the appropriate rate coefficients to balance
ionization against diffusion losses. This led to values of Ek = 3.0 ev.
E/N 1.1 x 10-l’ Vcm2/molecule and ki = 7 x in surprising agreement with the

above. With that EIN, an estimate of kd can be obtained from ref. 14, Fig. 8 by multiplying
the excitation coefficient (8 x
(1.5 x lo7 cm/sec) which gives kd = 1.2 x
cm2/molecule) by the drift velocity
also in fortuitously good agreement
with 1.1 x above. This excitation coefficient includes other, non-dissociating
processes, but since application of Corrigan's' larger dissociation cross section
would increase3otal excitation coefficient , its subsequent apportioning would not
lead to serious discrepancies. *
I
Thus, a very large atom production term of about 200 torrlsec is indicated by
all available estimates. In addition, there is the source term due to electron-ion
recombination at the wall, and due to the fast reaction €12' + Hp -+ H3+ + li. Each of
these can at most equal the ionization rate, so that their sum will contribute 10 to
20 torr/sec of atomic hydrogen. All other source terms are negligible.
Surface recombination provides the only comparably large loss term in the
discharge. For small y, the first order rate constant, k, = yC/d = 2 x 105y, and the
steady-state atom concentration, [HIss is approximately 1 x 10-3/y. This shows that
y must alwasy be relatively large, and that the diffusion controlled
ks = 5.8 D/r2 = 2 to 4 x lo4 sec-' may be applicable which leads to an [Illss
to 1 x lob2 torr, less than 1% mole fraction. Idhen y is lowered to
of 0.5
[HIss = 0.1 torr, and under either condition the halflife for its formation is very
short (3 x or 3 x lO-'?sec).
One is thus led to the interesting conclusion that the dissociation yield of
such H2 discharges is mainly controlled by surface recombination and that the surface
in the discharge must be moderately to highly efficient for atom recombination. More-
over, the effect of small amounts of added gases such as H20 or 02 in increasing the
yield must be through their influence on the surface efficiency, as they can not affect
the principal production term and as no homogeneous loss mechanisms of proper magnitude
are available. These conclusions are in general agreement with recent work by Goodyear
and Von Cngel" on radio frequency discharges at lower pressure and of different
geometry.
The very large body of experimental work on the "catalytic" production of H
when small amounts of Hgi) or 02 are added will not be reviewed here. The effect is
unquestionably real, i.e. factors of 10 or so in 11-atom yield when switching from
"dry" H2 to H2 containing 0.1 to 0.3% H20 can be demonstrated routinely and reversibly.
Rony and Hanson18 have recently questioned the existence of such an effect at a pressure
of 0.075 torr and have reviewed the general subject. Our early experiments do
not bear this out, as they show a large catalytic effect for added H20, but they do
indicate that the effect is less pronounced at pressures below 0.1 torr. This trend
can be expected if the above analysis is correct, because at low pressures, the production
terms are decreased and the loss terms increased. Moreover, for a given
discharge energy input, the surface should be hotter at low pressure, and this may
nullify the deactivation which is probably due to adsorbed water molecules. Several of
the relevant experiments are now under way in our laboratory by Dr. D. A. Parkes, using
Lyman-! absorption for the do-tream measurement of [HI.
111. 2. Oxygen Discharges
Electron collision cross sections have been calculated by Hake and Phelps19
from all available experimental data on electron drift velocity and other transport
coefficients. The cross sections for vibrational excitation are much smaller than in
Hp and that process can probably be neglected. Electronic excitation is described by
* It is Dr. phelps's view that the high E/N portion of the enelastic collision frequency
and the corresponding cross sections will have to be modified in view of recent work
by Fletcher and Haydon, Austral. J. Phys. 19, 615 (1966).

20
three processes with thresholds at 4.5, 8.0, and 9.7 ev. The latter two are larger
than the first, and should lead to dissociation, since the upper states have shallow
potential energy mimima at larger internuclear distances than the ground state. The
combined cross section for the 8.0 and 9.7 ev excitation shows a broad first peak of
1.1 x cm2 at 12 to 15 ev, then drops slightly and rises again to a second peak
in the 50 to 100 ev range. With the assumption of an average electron energy, Ek, of
3.0 ev the effective dissociation rate constant, kd, is about 2 x cm3 molecule-’
sec-l.
I
The corresponding ionization rate constant, ki, was calculated to be
1.3 x cm3 molecule-’ sec’l. Fitting the 02 data in a similar way to lip, Dr.
Phelps obtained E/N = 9 x
V cm2/molecule, Ek = 3.4 ev, kd = 1 x lo-’, and
k, = 1 x 10-l’ cm3 molecule-’ sec-l, in fair agreement with the above except for the
smaller ki, but k is strongly dependent on E/N near 10-15Vcm2/molecule, so that a
20% increase of ESN would lead to a tenfold increase of ki.
An patom production rate of about 200 torr/sec by dissociative electron
impact is thus indicated, and all other production terms are negligible by comparison.
Surface recombination of 0-atoms is kinetically similar to that of H-atoms except for
a lower diffusion coefficient and molecular velocity by a factor of 3. Thus, the
effective first order surface recombination rate constant is 6 x lo4 y sec-l for Emdl
r and 5 x 10’ 8ec-l for y approaching unity.
In pure 02, as in Np, there is another loss term which does not arise in H2.
The polyatomic ions Os+ and O4+ can react exothermically and rapidly via
01,’ + + 0 + 03 + 02, 03’ + 0 -+ 02+ + 02 to recombine 0-atoms. To obtain an upper limit
+
for this loss rate one may assume that 01, is a major ion, i.e. [O4+] * lf” ~ m - ~ ,
and that it is therefore regenerated by the bimolecular reaction 02 + 02 + 01,’. The
latter assumption requires that the lifetime of the unstabilized OI,+ collision complex
be equal to or longer than the collision time, 3 x sec, which seems excessively
long-for such a simple species. At its (unreasonable) maximum estimate, such a chain
process may recombine 0-atoms at twice the rate of @ + 0 -c O3+ + 02, i.e. with an
effective first order rate constant of 200 sec-l if both ion molecule reactions have
rate constants of cm3 molecule-l sec-’. Even so, it would not seriously limit
0-atom production, since the corresponding [O], = 0.5 torr = 50% mole fraction. Moreover,
Knewstubb, Dawson, and TicknerZ0 saw no Ob+ in their mass-spectrometric study
of dc 02 discharges at 0.4 torr, though the weaklv bound ion could have dissociated in
the large electric field at the sampling orifice as Schmidtz1 suggests for a in his
mass-spectrometric study of nitrogen ions. The above mechanism does have the desirable
property that it is easily quenched by small amounts of added gases such as N2 or H2
which are capable of transforming the oxygen ions into more stable ions such as NO+ or
EJO’, but it is unlikely to be important here.
The process 0- + 0 -c 02 + e is known to be-fast (k * 1.5 x and it should
follow the dissociative attachment step e 2 O2 + 0 + 0 which has a threshold of 4.5 ev
Insofar as 0 is formed mainly by this reaction ad
and a low maximum near 7 ev.
removed by its reverse, and the concentrations of electrons and ions are very much
lower than those of neutral species, these steps leave [O] unchanged.
If the above large excitation-dissociation rates are approximately correct, the
O-atom yield, as the H-atom yield in the preceding section, is principally controlled
by surface recombination in the discharge. The smallest calculated [Ol,, (y 1) is
about 0.04 torr, 4% mole fraction, considerably larger than experimental values for
pure 02. A possible explanation of this discrepancy may lie in the surface properties
immediately downstream from the discharge region. Active discharges have a sharp
boundary as shown by their light emission, because electron loss processes are Very
fast. The corresponding transition of the surface from a region where y is
i
near unity
to one where it is less than is likely to be more gradual, and would provide a
f

21
region in w!iich large, nighlv localized surface 10s; terns could qulcLlv reduce tile
atom concentration.
Experimentally, 1ary.e catalytic effects by :;?, SO, or li2 in tile production of
0-atoms in microwave ciisciiargcs ~iave ~cerl reported.22 Very ptire oxygen qave only u.6~
atoms (still ~ O W ~ yields K oi 0.3';; WCKC latcr obtained), !>ut small adtiitions (U.01 to
0.u5;:) of Sz, >;20, or !;o produced +atoms at tile rate of 4u to 45 per added ;:, and
similar additions of 112 produced 160 to 200 +atoms per added 112. Ih terms of the
!'resent interpretation, tile large catalytic cffrct may he understandable for ti2 additions
as due to t!20 wall effects, !,ut less so for nitrogen compounds which sliould not
be StKOIlGlY adsorlied at tile surface. (:onceivably, XO+ ir ::02+, stron): Lewis acids, my
be involved in poisoiiiny, the surface. Thus, our understanding of 02 discharzes is still
in an unsatisf actory state. Furthcr experiinents are required in wliich particular
attention should he given to tile condition and cliaracterization of the surface as vel1
as to the imnedinte downstream rccion.
111. 4. SitroRcn Disc!iarges.
The great complexity of "active nitro1:en" is probably due to its larger cross
sections for vibrational excitation anci to tile existcncc of metastable electronically
excited states helot) tlie dissociation l i m i t of ground-state 1iz. Consequently, extensive
vibrational excitation persists for times much longer than those spent in tlie discharge
zone, and chemiionizatioll is observed in regions such as tlie "pink glow" well downstream
of the discharge. The absence of the lowest triplet state, A,3Z:, in active nitrogeng3
contaiiiing '.';-atoms indicates that tliesc excited molecules are very efficiently quenched
by S, and that vibrationally liiglily excited ground-s tate molecules arc the principal
carriers of excitation to tlie dowiistreai~! reqioii. Engelliardt, Phelps, and Riskz4 have
determined tlie relevant elastic arid inelastic electron collision cross sections. Some
of the electronically excited states above tne dissociation limit do not lead to pre-
dissociation, and thcrefore only the state with tlireshold energy of 14V was used in
tile estimate of dissociation. Assuming an average electron energy. = 3 ev, and a
maxwellian distribution, one obtains an effective dissociation rate constant, kd, of
3 x lo-'' (bo torr/sec) and a correspondinz ionization rate constant, ki, of 6 x lU-l'.
'The latter is larzer (b torr/sec) than the corresponding ambipolar diffusion loss term
(13.5 to 1 torr/sec). lilt more realistic calculation by Dr. Phelps which simultaneously
fits ck, L/:i, and tlie known cross sections to make the ambipolar diffusion loss equalthe
rate of ionization gave ck = 2.2 ev, E/?: = 1.2 x cm2/molccule. kd = 3 x 10 "
(6 torr/sec), and ki = 3 x This dissociation rate is very much lower than that
of 1iZ or O2 and properly reflects the tiifficulty of producing extensive dissociation of
Hz in glow discharxes. ?io other source terms of comparable magnitude are available.
The principal loss processes include atom recombination at the surface which can be set
equal to those of oxygen, because the molecul K velocities are similar. The catalytic
a atom loss nechanicm by E;,++ + X + i

1
\
22 .
Experimentally, there is abundant evidence for "catalytic" effects as summarized
by Young et alZ8 who studied the effectiveness of 02, NO, and SFg added either
before or immediately after a nicrowave discliarge in highly purified, flowing Xz.
\
All three gases were "catalytic" when added before, but only NO after the discharge, ,
and SF6 added before "produced" 230 !

i
’ understanding:
23
(a) Electron impact ionization and aissociation rates of reactants
1 and other major species; (b) Surface recombination ratcs in the discharge: (c)
Ion-molecule reaction rates involving major neutral species; (d) neutral-neutral
reaction rates and their tcmpcrature dependences in and out of the disciiarge. ‘[uch of
) the necessa information for (a), (c), and (d) is becoming available for many systems.
The role OBurface, however, is little understood for electrically neutral reactants,
1
I
, even less so for charged species, and seems to be the present bottleneck.
\
i 1.
I
. \
2.
3.
I 4-
5.
6.
’ 7.
1 8.
Y.
10.
1 11.
’ 12.
’, 13.
1; 14.
15.
I 16.
17.
j 18.
19.
20.
21.
22.
23.
, 24.
1 25.
i 26.
1 27.
28.
, 29.
Acknowledgement :
The author would like to thank UK. A. V. Phelps for many interesting discussions
and for calculations of important parameters.
References
H. S. W. Massey and E. H. S. Burhop, Electronic and Ionic Impact Phenomena, Oxford
University Press, 1952.
E. N. McDanel, Collision Phenomena in Ionized Cases, John Wiley and Sons, N.Y. 1964.
P. Llewellyn-Jones, The Glow Discliargc, ?lethuen and Co., London, 1966.
s. c. Brown, Introduction to Electrical Discharees in Gases, John Wiley and Sons,
N.Y. 1966.
L. Goldstein, Advances in Electronics and Electron Physics, Vol. 7, 399-503 (1955).
S. C. Brown, Handhuch der Physik, Vol. 22, 531-575 (1956).
!4- A. Uiondi, Advances in tlectronics, Vol. 18, Academic Press, Inc., N.Y.
‘-4. ii. Kasner and ?4. A. Uiondi, Phys. Kev. 137, A317 (1965).
B. li. Mahan and J. C. Person, J. Cliem. Phys. 40, 392 (1964).
T. S. Carlton and B. li. Xahan, J. Chem. Piiys. a, 3683 (1964).
1963, p.67.
D. Rapp and U. 1). Urif,lia, J. Chem. Phvs. a,
1480 (1965).
F. C. Fehsenfeld, E. t. Ferguson, and A. L. Schmeltekopf, J. Chem. Phys. 44, 1844
(1966).
Y.
A.
Tanaka, F. R. Innes, A.
G. Lngelhardt and A. V.
S. Jursa, and ?I. :

24
Ion-Molecule Reactions in Electric Discharges
3. L. Franklin, P. K. Ghosh and Stanley Studniary
Rice University, Houston, Texas I
Before the development of good vacuum equipment and techniques, mass spectrometrists
observed many ions occurring at masses well above those of t,he molecules admitted
to the instrument. These interferred with the main interests of the experimenters
at that time, although in a few instances it was recognized that reactions
were occurring between primary ions and molecules. When good vacuum facilities became
available in the early 1930s they were adapted to mass spectrometry and for a
number oi years every effort was made to maintain pressures well below torr ,
in order to avoid collisions of primary ions with neutrals. However, in 1940, Mann,
Hustrulid and Tate (l), in studying water, observed an ion at mass 19, and by vary-
I
(1) Mann, M.M., Hustrulid, A. and Tate, J. T., -. Rev., 58, 340 (1940)
ing the pressure in their instrument, concluded that it was H30+, probably formed
by the reaction: + +
H20 + H20 + H 0 + OH
3
During the middle 50s, several groups of mass spectrometrists became curious as to
the results that would be observed if source pressures were raised sufficiently to
allow a few collisions to occur between ions and molecules. Several ions occurring
at masses above those of the parent ion were observed. Because of this, a number
of systematic studies were made and increasing interest and emphasis on reactions
of ions with neutral molecules has developed, until at present some 50 to 100 papers
per year are published on this subject alone.
Perhaps one of the strongest reasons for the interest in ion-molecule reactions
is the fact that ions of unusual and unsuspected composition were observed. Most
fascinating of these has been CH5+, which was completely unexpected, but which,
nevertheless, is now well established as a stable ioc. This ion was first announced
by Tal'roze and Lyumbimova (2), but it was also reported shortly after the Russians
(2)
Tal'roze, V. L. and Lyumbimova, A. K., Dokl. Akad. Nauk SSSR, 86, 909 (1952)
by several groups in this country (3,4,5,9).Studies of secondary ions from a large
number of molecules followed.
(3) Field, F. H., Franklin, J. L. and Lampe, F. W., 2. Amer. Chem. %., 79, 2419
(1957)
(4) Stevenson, D. P.and Schissler, D. O., J. Chem. PQ., 2, 1353 (1955)
(9) Schissler, D. 0. and Stevenson, D. P., 2. Chem. Phys.,-24, 926 (1956)
(5) Meisels, G. G., Hamill, W. H. and Williams, R. R., Jr., J. E m . -., 2, 79c
(1956)
The earlier studies of secondary ions were largely limited to ions of masses
greater than that of the parent ion, but subsequent studies have shown that many
secondaries of lower mass also occur. In most instances, the fact that the ion re-
sulted from collision rather than from impurities was demonstrated by varying the
pressure in the ion source. If the ion intensity increased as a second power of
the pressure, the ion clearly resulted from a collision.
Having established that an ion was indeed the result of a collision, it was
obviously of interest to ascertain the primary ion precursor of the secondary ion.
TWO methods have usually been used for this purpose. The method most often used
was to measure the appearance potential of the secondary ion and compare it to
appearance potentials of various primary ions occurring in the system in question.
-
-
-
-
,'
,
1

Obviously, the secondary ion must have the same appearance potential as its precursor,
and where reasonable agreement of primary and secondary ion appearance potentials was
found the identities of reactant and product were established. In some instances, it
has been possible to reduce the electron energy to the point where only one or two
Primary ions were present. Under these conditions, it is often possible by comparing
intensities of secondary ions with the disappearance of primaries to identify unequivocally
the precursors of the various secondaries. Obviously, boqh of these
techniques suffer from certain difficulties. Where the primary ion spectrum is sufficiently
complicated, or where the appearance potentials of several primary ions are
sufficiently close together, it is very difficult and often impossible to determine
the precursor of a secondary ion satisfactorily.
Further, comparison of primary and
secondary appearance potentials usually serves to identify only the precursor of
lowest appearance potential. Possible precursors of higher appearance potential
are obscured and can only be detected by other means. Of course, it is also true
that one cannot always simplify the primary spectrum sufficiently by reducing the
electron energy to permit satisfactory identification of the precursor. As a result,
in recent years a few mass spectrometrists following Lindholm (6) have employed a
primary mass sorter to select the primary ion which is then injected into the gas
(6) For a survey of Lindholm's work see E. Lindholm, "Ion-Molecule Reactions in the
Gas Phase," Advances in Chemistry Series 58, American Chemical Society, Washington,
D. C., 1966, pl.
with which reaction is desired.
Mass spectrometer ion sources are quite small chemical reactors. It can easily
be shown that the reaction time of an ion in the source will normally be in the
order of a microsecond, and unless the pressure in the source is well over 100 torr
only a small fraction of the ions can undergo collision. It early became apparent
that where secondary ions were observed they must, in most instances, have resulted
at almost every collision of ion and neutral. Further, in many instances, the
heats of formation of ions and neutrals were well established, and in all such
instances it was possible to show that the reactions were exothermic. Indeed, it
would be impossible under most circumstances to observe a reaction to form secondary
ions if the reaction were endothermic. Such endothermicity would appear as activa-
tion energy, which would greatly reduce the probability of reaction when the ion
and molecule collide. (Later on we mention certain conditions in which this rule
is violated, but under most circumstances it holds rigorously.) This rule is so
seldom violated that it can be used as a means of helping to eliminate possible
precursors of a secondary ion.
It was mentioned above that the secondary ions first observed occurred at masses
above those of the parent. However, there was always an expectation that ions of
lower mass might also be formed in collision reactions and as the pressure that
could be tolerated in the ion source was increased it became apparent that such was
indeed the case. For example, it was observed by Munson, Franklin and Field (7)
(7) M.S.B.Munson, J. L. Franklin and F. H. Field, J. Phys. Chem. 68, 3098 (1964)
that the C H + ion from ethane and the C3HTf ion from propane, both present in the
primary sp$c?rum+ increased with increasing pressure and indeed, in the case of
propane the C H ion increased from a very small proportion of the primary spectrum
3 7
until it represented some 70% of all the ions present. Many secondary ions of mass
less than the parent did not show such spectacular increases and other techniques
were sought to establish these. One method of especial interest was that due to
He employed electrons having energies too small to ionize in the
Cermak. (a),.
(8)
V. Cermk and 2. Herman, Collection Czechoslov. Chem. Commun. 2, 406 (1962)
ionization chamber. However, he employed a relatively high variable potential
between the ion chamber and the trap anode, so that the electrons were accelerated
-
-

I
26
in the anode region. Some of these ionized molecules present in that region and the
ions were repelled by the potential on the anode and drifted back into the ionization
chamber. These primary ions could not be themselves collected, but new second-
ary ions could be unequivocally identified.
Further, by varying the potential on
the anode, it was possible to obtain an appearance potential for the secondary ions
and from this to deduce the identity of its precursor.
A large number of reactions of ions with neutrals have now been identified and
typical examples of the various classes of such bimolecular reactibns are given in
tables 1 through 7.
Table 1
Atom Transfer Reactions
+
+ Kr + D2
+ Xe + CH4
+
N+ + D2
+
-+ KrD+ + D (9)
+ + XeH + Cis3 (10)
+ -+ N2D + D (9,11)
+ C3H5 (12)
5+
+ 0 + N2
+ 0 + C02
2+
-+ NO + N (14,15)
-+ 0 + CO (16)
+ CH3 (17)'
CH4 + C3H6 + CH +
D20 + nC4HI0 --f HD 0 + C4H9 (13)
+ 2+
I + cH31
-+ I2
(9) Schissler, D. 0. and Stevenson, D. P., J. Chem. Phys.,24, 926 (1956)
(Ip' u>ld, F. H., Franklin, 3. L., 2. &. *m. E. 83, 4509 (1961)
(11)-Stevenson, D. P., J. a. %m.,6l, 1453 (1957)
(12) Frankevich, E. L. and Tal'roze, V. L., Dokl. Akad. Nauk SSSRZ, 1174 (1958)
(13) Lampe, F. W., Field, F. H. and Franklin, J. L., J. 3. Km. E., 79, 6132
- (1957)
(14) Potter, R. F., J. E m . Phys., 23, 2462 (1955)
s=
(15) Fehsenfeld, F. C., SchmeltekopfTA. L. and Ferguson, E. E., Planet. Space Science
13, 219 (1965)
(16) zhsenfeld, F. C., Ferguaon, E. E. and Schmeltekopf, A. L., 2. Chem. Phys. 2,
3022 (1966)
(17) Pottie, R. F., Barker, R. and Hamill, W. H., Radiation Research 2, 664 (1940)
(18)
(19)
(20)
Table 2
Positive Atomic Ion Transfer Reactions
0 ++ H2 + H O++ 0 (18)
H2 2+ + O2 +€SO> + H (9,11) .
HI++ CH31 + CH I + + H (19)
33
C2H: + D20 +€ID 0 + C2H5
+ 3
O2 + N +NO + 0 (21)
0; + C2H2 --t C2H20+ + 0
Hutchinson, D. A., Paper presented at American Chemical Society Meeting,
Minneapolis, Minnesota, September, 1955
Pottie, R. F., Barker, R. and Hamill, W. H., Radiation Research 10, 664 (1959)
Lampe, F. W., Field, F. H. and Franklin, J. L., J. Amer. Chem. s., 2, 6132
(1957)
(21) Golden, P. D., Schmeltekopf, A. L., Fehsenfeld, F. C., Schiff, H. I., and 1
-
Ferguson, E. E., 2. Chem. Phys. 44, 4095 (1966)
(22)
(20)
c
1
\
!I
4
I\
\
,

27
(22) Franklin, J. L., Munson, M.S.B., Tenth Symposium (International) on Combustion,
The Combustion Institute, 1965, p. 561
(23)
(24)
(25)
(26)
(27)
Table 3
Symmetrical Transfer Reactions
+
H; + H2 i H, + H (23,24,11)
CH + CH4 i CH' + CH3 (2,3,4,5) i
4+ 5+
H 0 + H20 +H30 + OH (1)
2
HC1+ + +
HC1 + H2C1 + H (9)
+
CH202+ + CH202 4 CH 0 + HC02 (22)
NH;+ NH3 +NH4 3+2 + NH2 (25,26,27)
I2 + I2 +I + I (28)
3
Eyring, H., Hirschfelder, J. 0. and Taylor, H. S., 2. Chem. Phys. 3, 479 (1936)
Stevenson, D. P. and Schissler, D. O., J. %m. Phys.23, 1353 (1953)
Lampe, F. W. and Field, F. H., Tetrahedron 1, 189 (l"3)
Dorfman, L. M. and Noble, P. C., J. Phys. CEem. 63, 980 (1959)
Derwish, G. A. W., Galli, A., Giardini-G-idoni, A., and Volpi, G. G., J.
- Chem. Phys. 3, 1599 (1963)
and Harkness, R. W., Phys. Rev. 2, 784 (1928)
(28) Hogness, T.
Table 4
~
-
H- and H,- Transfer Reactions
+ L +
C2H5+
C H
+ C3H8
+ C3H8
+ C2H6 + C H
7+
+ C H + C3H7
(29)
(29)
4+ 5+
CH3 + C2H6 + C2H5 + CH4 (34)
+ C3H5 + neo-C 5 H 12 + tC5H:1 + '3f6 (34)
+ C H + C3H8 + C H + C3H6 (29)
2 6
24 C H + iC4H10 -+ C H + + iC4H8 (29)
3 6 3 8
+"
(29) M.S.B.Munson, J. L. Franklin and F. H. Field, J. Phys. Chem. 68, 3098 (1964)
Table 5
Some+$eactions of Excited Ions
+ C2H6 + C2Hl+ + C2H5 (29)
C2H6
H2+* + He + HeH + H (30)
N + N2
O2 2+* + O2
+ N -I- + N (21,23,33)
+ 0: + 0 (22,31)
(30) H. von Koch and L. Friedman, J. Chern. Phys. 38, 115 (1963)
(31) R. K. Curran, 2. Chem. Phys. 38, 2974 (1963)-
(32) M.S.B.Munson, F. H. Field and?. L. Franklin, 2. Chem. Phys. 37, 1790 (1962) -
(33) M. Saporoschenko, Phys. Rev. 111, 1550 (1958)
(34) Field, F. H. and Lampe, F. W., J. Amer. Chem. z., 2, 5587 (1958)

C H +
28
Table 6
Charge Transfer Reactions
+
0 +02 -+ 02+ + 0 (35)
N2 + + O2 + N2 + 02+ (36)
+
2+
C2H4 --f C H +
Xe + C2H4
Ar + + C2H4
+
Ar + CH4 4
C2H2 (37)
+
c
4+
C H + Xe (38)
I lH4' + H: + Xe (38)
C2H2+ + H2 + Ar (38)
C:H; + H + A r (38)
CH? + H + Ar (5,39)
F++ CO -+ CO + F (40)
I, C+ + 0 + F (40)
(35) F. C. Fehsenfeld, P. D. Goldan, A. L. Schmeltekopf and E. E. Ferguson, Planet.
Space Sci. 2, 579 (1965)
(36)
F. C. Fehsenfeld, A. L. Schmeltekopf and E. E. Ferguson, Planet. Space Sci. 13,; -
919 (1965)
(37)
(38)
(39)
F. H.
J. L.
G. G.
Field, 2. fi. %m. e. 83, 1523 (1961)
Franklin and F. H. Field, 2. Am. Chem. E. 83, 3555 (1961)
Meisels, W. H. Hamill and R. R. Williams, Jr; 2. Phys. Chem:Z, 1456
(1957)
(40) E. Lindholm, Arkiv Fysik - 8, 433 (1954)
CH4 + O2 -+ CH30 + OH (22)
+
+
Table 7
Xe + C H ~ -+ Xem; + H (10)
t
I
I
1

Tables 1-6 present examples of relatively simple reactions. However, reactions
involving quite profound changes in structure and bond reorganizations have been
observed by a number of investigators. Table 7 shows several of these, which are
given as typical examples. Certain of the reactions shown result in more than one
set of products, all of which apparently occur at the same appearance potential
and thus involve the same precursor. Of course, the intensities of the product
ions will, in most instances, not be the same. One would expect tha,t reactions
of this kind would involve the formation of a relatively stable complex, which
breaks up in ways dictated by the energy content of the complex. Unfortunately,
no quantitative treatment of the break up of the complex has yet been published,
so it is not now possible to predict the ratios of product ions where more than
one product arises from a single reactant. It has, however, been observed by Lampe,
et al. (41) that the ratios of secondary ions from a collision complex will often
be very similar to those of the same fragment ions in the primary mass spectrum
of a compound having t e Sam$ composition as the comp ex. Thus, they pointed out
9 4
that the ratio of C H /C H in the reaction of C2H4 with C H was about the
35. 4 7
same as that observed in the mass spectra of the butenes. Thg fiterature contains
(41) F. W. Lampe, J. L. Franklin and F. H. Field, "Progress in Reaction Kinetics,"
Val. 1, Pergamon'Press, New York, 1961, p. 69.
only a few examples of simple condensation reactions. This is not surprising in view
of the fact that every complex is formed with enough energy to break up in either
the forward or reverse direction. Since most complexes can break up very rapidly
they will generally do so in a time short compared to that required to collect the
ion. In a few instances such complexes have been observed to survive long enough
to be measured. One example of this is given in Table 7. It might be mentioned
that several apparently long lived complexes were observed by Field (37) in his
study of ethylene, but these in all cases turned out to depend upon the third or
higher power of the pressure, and thus were in fact complexes that had been
stabilized by collision or had resulted from the decomposition of a complex of
higher molecular weight.
Although the preceding discussion has largely been devoted to ions formed as
bimolecular reaction products, in the last few years, many examples of ions formed
with much higher pressure dependence have been observed. One of the earliest studies
carried out at such elevated pressures, i.e., above 100 microns, was the study of
Field (37) of ion-molecule reactions in ethylene. In this study he observed ions
with pressure dependencies as high as about 6, although those exhibiting the highest
pressure dependence were of such low intensity that the nature of the reactions
involved could not be ascertained. Indeed, above about 3rd order the method of
appearance potentials becomes completely useless, and the identification of precursors
to a given product becomes very tenuous indeed. To form ions at high pressures
in a region from which they can be collected it is usually necessary to employ
electrons of several hundred volts rather than the usual 60-70 volts employed in
most mass spectrometric problems. Indeed, because of this problem, Kebarle and
Hogg (45) have employed alpha particles of high energy to provide primary ions.
(45) P. Kebarle and A. M. Hogg, J. &m. a. 42, 668 (1965); 2, - 449 (1965)
When ions are formed by high energy massive particles it is possible to operate
at much higher pressures and Kebarle (45), Wexler et al. (46) and others have
(46)
S. Wexler, Assa Lifshitz and A. Guattrochi, "Ion-Molecule Reactions in the
Gas phase," Adv. in Chem. Series, Amer. Chem. SOC., Washington, D. C. 1966
p. 193
studied the ions formed at pressures up to about one atmosphere. Naturally, they
have observed ions with a very high pressure dependence, although they have not been
able to establish the order of reaction. With such a system, Kebarle et al. (49,50)
-

-_
30
have observed ions having the general formula, H+(H O)n, with n varying from 1 to 7,
and Wexler, et al. (46) have observed polymer ions Prom acetylene having up to 12
carbon atoms.
Recently, studies have been carried out in mass spectrometry sources at pressures
up to several torr. Under these conditions, any primary ions formed will undergo
many collisions and will have ample opportunity to react if they are capable of
doing so. Field and Munson (47,48) have taken advantage of this to carry out some
very intereciting studies of reactions of highef order. pey observeh that in very
pure methane, the principal secondary ions CH5 and C2H5 reached a plateau and re-
(47) M.S.B.Munson and F. H. Field, 2. &. =m. %. 88, 2621 (1966)
(48) F. ti. Field, M. S. B. Munson and D. A. Becker, "Ion Molecule Reactions in
the Gas Phase," Adv. in Chem. Series, Amer. Chem. SOC., Washington, D. C.
(49)
(1966), p. 167
P. Kebarle and A. M. Hogg, 2. s m . m. 42, 798 (1965)
(50). A. M. Hogg and P. Kebarle, J. Chem. Phys. 43, 498 (1965)
mained constant with increases in pressure beyond about 0.5 torr. They observed
also that if there were any small amount of impurities in the methane the intensities
of these ions passed through a maximum around a few .loth8 torr and then declined
steadily with further increases in pressure. The primary ions had very small probability
of colliding with anything but methane and consequently the disappearance
of the secondaries must be due to their reaction with the impurities, since it is
demonstrated that they did not react with methane itself. It was a simple step
from this to the addition of small amounts of a variety of materials to the methane
plasma, with results that proved extremely interesting. When, for example, smell
amounts of long chain paraffin hydrocarbons, such as dodecane, were added to the
methane plasma, the spectrum of ions from the high molecular weight paraffin was
quite different from that obtained by electron impact. Such high molecular weight
paraffins give only small intensities of ions above about the C5 range under
eleqtron impact. However, in the methane plasma, ions of the general composition
C H formed at each carbon number from the parent down to C4. No doubt ions
o? tailer mess are also formed, but these are not observable because of the interference
from secondary and ternary ions from methane. Further, the largest of thfse
ions is the one corresponding to the parent molecules; i.e., with dodecane,C12H25 .
Field and Munson have studied a number of compounds by this method, and in
many instances have obtained profound changes in the mas spectrum, produced by
"Chemical Ionization," the term which they have given the processes. (47,48)
They have concluded that in the case of the methane the principal reactions are
probably as follows:
CH5 + + CnH2w2 + CH4 + H2 +'Cn+H + ,
2#1 .
-
+
4 C H + CnH2#l
2 6
Although the previous discussion has been devoted entirely to reactions of positive
ions, negative ions are also known to undergo reactions on collision. Relatively
few of these reactions have been studied, however, largely 'because negative
ions present some rather serious difficulties to the investigator.
of negative ions, however, have been carried out by Melton, Henglein and others,
and several typical reactions are given in table 9.
-
c
Some reactions

Table 8
Some Ions Formed by Processes of Order Higher than 2
Rea c tan t Ionic Product Reference
H2°
m3
H+(H,O), (1

J
32 ',
1
A number of investigators have subsequently studied this chemi-ionization reaction of\
the rare gas ions and have confirmed and extended Hornbeck and Molnar's observation,
It has now been observed that all of the rare gases react with each other to form
.the hetero-nuclear diatomic ions (57). In addition, a number of ionic compounds of ,
the rare gases with nitrogen, CO, 02, methane, acetylene and others have been report-'
ed. Chemi-ionization reactions are not necessarily limited to rare gases, however. , !
Chemi-ionization products of excited mercury atoms with a number of compounds have
been observed (59) and nitrogen and CO are known to und r o chemi-ionization with
their own ground state species, forming respectively N -? :nd C20; '(32).
4
1
I
(59) P. Cermak and 2. Herman, Tenth Annual Conference on Mass Spectrometry, New
I
,
Orleans, La., 1962, p. 358. I
,
It is natural that when both reactants and products are readily measurable
c nsiderat'on should early be given to the fate of the reaction. With the reaction
4? + S, if M is much larger than A , the feactior rate is psuedo first order'
A + M -%
and the equation expressing the concentration of A and B is as follows:
- B+ =
AO
+
-kMt
1-e
-
-1
A Llot of B+ against M will yield a straight line whose slope is kt. If a
At relatively low pressures, (a few microns), i
time is thus:
B+
- =
A' + B+
rate constants for second order ion-molecule reactions are given in table 11. It
will be observed that many 2 them are in the order of 10 cc/molecule/sec. How-
€3
ever, values as small as 10 cc/mo_l culelsec have been reported for some reactions.
6 The values in the neighborhood of 10 cc/molecule/sec represent reactions that must
occur at essentially every collision in that they have cross sectionslSonsiderably
larger than ordinary collision cross sections. The values around 10 represent
or very small.
must not involve appreciable energy of activation. Stevenson and Schissler (4,11)
have confirmed this experimentally for a few reactions and the very fact that a
kMt
As was mentioned before, ion-molecule reactions to be observable ,
~ ~~
I

33
velocity of the reacting partn rs is sufficie tl y great. Giese and Maier (60) have
f shown that for the reaction Ar + CO 4 Ar + C + 0, which is endothermic by 6.62
+
and 6.44 e\’ for the 2p state of Ar , the threshold for the
“co
Their measured values were in agreement with this relation.
(60) C. F. Giese and W. B. Plaier, J. X m . a. 39, 197 (1963)
(61)
(62)
(63)
React ion
1
112
Table 11
Some Second-Order Rate Constants
10 10 k, cc/molecule sec Reference
D2 -3 D3 + D 14.5 4,11
+
CH4 + CH4 + CH
5+ + m3
10 61,62,63
H20 + H20 --f H30 + OH 12.7 1,2,13
+ +
C H + C H + C H +CH4 14.5 9
2 6 3+5
2+3
+ C H + ~ CH o + n 0.126 22
O2 + 33
CH4 + O2 + CH30 + OH
0.257 22
V. L. Tal’roze and E. L. Frankevich, J. Phys. Chem. USSR 34, 1275 (1960)
C. W. Hand and H. von Weyssenhoff, Can. J. Chem., 42, 195, 2385 (1964)
J. L. Franklin, Y.
2353 (1966)
Wada, P. Natalis and P. M. Hierr J. Phys. Chem. 70, -
All reaction rate studies have not been limited to extremely low pressures.
When the pressure in the ion source becomes sufficiently great the reactions follow
the psuedo first order rate laws over rather wide ranges of pressure, unless subse-
quent reactions interfere. Thus, Field et al. (51) found the disappearance of CH4
in methane to obey first law kinetics over a pressure range of about 0.1 to 400 Torr.
However, as the pressure is raised, in certain systems, reactions of a higher order
do occur. Field, (37) studying ethylene, has reported some reactions having apparent
orders as high as about 6. These, of course, are not truly 6th order reactions, but
simply represent a succession of perhaps five secondary reactions which show dependence
upon the 6th power of the pressure. Rates of reactions of such high order do
not yield satisfactory rate constants, but it has been possible to determine rate
constants for reactions of 3rd order, and Field and several others have made approximate
measurements of the rate constants for such third order processes. As is the
case with the second order processes, these reaction rates are relatively high. For
example, in ethylape eld deter ined several third order rate constants to be
F3 2 -1
in the order of molecule- sec .
Although most of the measurements of rate constants in the literature were
obtained with the source operating in a continuous mode employing a variation in
pressure to establish the rate, some studies have been made in which retention time
in the source was varied.
Such measurements were originally made by Tal’roze and
Frankevitch (61), but subsequent measurements have been made by Hand and von Weissenhoff
(62) and Franklin, Natalis, Wada, and Hierl (63). In order to obtain a satisfactory
variation of time, a pulsed mode is employed. In such an operation a pulse
of electrons is fired through the gas. After it is stopped, the resulting ions
can be retained in the source for a controlled period of time and then rapidly
extracted by a pulse of high energy. By varying the delay time, the time of retention
in the source is varied, and rate constants determined in the manner more usually
employed by chemists in rate studies. Results obtained in this way generally agree
-
I
+

.
34
rather well with those obtained by the pressure method, although some differences
have been observed. One difference that may be of significance is that the ions in
the pulsed mode will generally have approximately thermal energies, whereas those
reacting in the continuous mode will have variable energies, depending upon the point
of their reaction in travelling from the electron beam to the exit slit.
The question of the effect of ionic energy on reaction rate has been one of
considerable interest and the subject of a number of investigations, both theoretical
and experimental. In their early work on ion-molecule reactions, hranklin, Field and
Lampe (3) observed that when they varied the field strength in the ion source in
order to vary the retention time, the rate constants that they calculated for their
ion-molecule reactions varied considerably. In general, they seemed to drop as
the field strength increased, suggesting that the reaction rate constant decreased
with the relative velocity of ions and neutrals. Other investigators have made
similar observations. However, it appears that not all reactions show such reduction
in rate constants with increasing relative velocity. Attempts to explain this have
been made by a number of investigators. Field, Franklin and Lampe (3) attempted
to obtain the theoretical relations based upon a balance of polarization and
centrifugal forces. Gioumousis and Stevenson (64) derived a more precise expression
(64) Gioumousis, G. and Stevenson, D. p., 2. Chem. Phys.,g, 294 (1958)
(65) P. Lannevin. Ann. Chim. Phvs. 5. 245 (1905)
for the collision rate based upon LaiTgevin's(65) treatment for polarizable systems.
Gioumousis and Stevenson found the collision cross section to be:
where a is the polarizability of the neutral, v
i
is the velocity of the ion,
lJ is the reduced mass and e the charge on the electron. Since
1'
(I will vary as Further, since k = uv
and thus is independent of velocity or energy. This, unfortunately, did not agree
with the observed rate behavior of a number of reactions, although it appears to
hold for some. In fact, Field, et al. (3) studied the effect of field strength upon
rate constant and found that for several reactions k decreased with increasing field
strength. Hamill and his associates (66,67) have shown that ion-molecule reactions
cannot be accurately treated as involving point particles. By considering the de-
(66) N. Boelrijk and w. H. Hamill, J. &. %m. =. 84, 730, (1962)
(67) D. A. Kubose and W. H. Hamill, J. &. X m . e. 85, 125 (1963) -
formable neutral to exhibit a hard core to high energy collisions while being deformable
in low energy collisions, these workers showed that for small ion energies
u obeped the Gioumousis-Stevenson relation (equation 5), but for large energy
uk E which agrees with experimental results. Theard and Hamill (68) and Moran
and ffamill (69). have extended their treatment to ion-molecule reactions involving
(68) L. P. Theard and W. H. Hamill, J. Am. -m. E. 84, 1134 (1962)
(69) T. F. Moran and W. H. Hamill, J. Chem. Phys. 39, z13 (1963)
neutrals with permanent dipoles. They showed that at law relative velocities a
=ne,
cross section for the ion-permanent dipole interaction, ~~
-
Et
(4)
(5)

35
must be added to the Langevin cross section. Here is dipole moment and Etis the
translational energy of the reacting system in center of mass coordinates.
NO attempt will be made here to review all of the studies of the effect of energy
upon the rates of ion-molecule reactions. However, several investigators who have
made especially important contributions should be mentioned. In addition to the
work of Hamill discussed above, important studies have been made by Futrell and
Abramson (70), Giese and Maier (71), Friedman (72,73) and Light and Horrocks(74).
- -- --
(70) J. H. Futrell and F. P. Abramson, "Ion-Molecule Reactions in the Gas Phase,"
Adv. in Chem. Series 58, Amer. Chem. SOC., 1966, p. 107.
(71) C. F. Giese and W. B.xaier, J. Chem. Phys. 39, - 197, 739 (1963)
(72) T. F. Moran and L. Friedman, J. Chem. Phys. 39, 2491 (1963); 62, 2391 (1965) - -
(73) F. S. Klein and L. Friedman, J. Chem. Phys. 4l, 1789 (1964)
(74) J. C. Light and J. Horrocks, Proc. Phys. 527 (1964)
It should be pointed out that if the relative velocities of ions and neutrals
become sufficiently high other reactions begin to OCCUL as a result of the different
forces coming into play. Thus a fast moving ion passing a molecule with sufficient
velocity may simply strip off a peripheral atom, leaving the partially denuded entity
behind. Such stripping reactions have been studied by Henglein (75) and Koski
(76), who found that they obey quite different rules from those above. It thus
(75) A. Henglein, "Ion Molecule Reactions in the Gas Phase," &. in Chem. Series
- 58, Amer, Chem. SOC., Washington, D. C.,' p. 63
-
(76) M. A. Berta, B. Y. Ellis and W. S. Koski, ibid., p. 80
appears that a complete theory of the rates of ion-molecule reactions has not been
developed, but there is little doubt that to a first approximation the equation of
Gioumousis and Stevenson givsfairly good results. It is also true that the
reactions tend to be quite fast, and this becomes a matter of overriding importance
in many processes involving ions. Thus is has been shown by Lampe (77,78) and by
Futrell (79) that ion-molecule reactions are the controlling factor in certain
(77) F. W. Lampe, Radiation Research lo, 671 (1959)
(78)
(79) J. H. Futrell, J. &I. -m.
82, 1551 (1960)
=E
*. 8l, 5921 (1959)
-
F. W. Lampe, 2. &. k m . z.,
-
radiation induced processes involving paraffin and olefin hydrocarbons.
An electric discharge of course involves ions and electrons and the ions present
can be sampled and analyzed by mass spectrometry. Such studies have been undertaken
by a number of investigators and some progress is being made toward understanding
the chemical behavior of ions in various discharges. The problem is complicated by
the fact that there are several kinds of electric discharge, each of which has its
own physical and chemical characteristics. Of these, the type most often studied is
the direct current glow discharge, but corona and high frequency and micro-wave
discharges have also received some attention.
In order to analyze the ionic content of a discharge it is necessary to trans-
fer the ions from the discharge into the mass analyzer. While this is not a par-
ticularly serious problem at pressures below 10 microns, the difficulty becomes
more acute as the pressure increases. This is attributable to the fact that elec-
trons and ions diffuse to the walls at different rates, so that an electric gradient
is established which alters the distribution of energy of the various charged species.
Ordinarily a sheath of ions is formed at any surface, including that of a sampling
probe and ions or electrons must have, or be given enough energy to pass through
this sheath in order to be sampled. However, if the energy is sufficiently high
some ions may be decomposed by collision. The result then is that there is often
-

36
considerable undertainty as to the quantitative correspondence of the ion distribu-
tion reported by the mass spectrometer with the actual distribution in the discharge,
Further, while there is little doubt that the ions observed were actually present,
there is always some question as to the presence of ions that might be expected, but
that are not observed.
An added complication that must be taken into account, especially in glow dis-
charges, is that different portions of the discharge have different characteristics,
Thus the negative glow and positive column have quite different el'ectric fields, the
ions and electrons present have different energies and the distribution of ions in
the two regions is different.
In spite of these reservations, considerable information has been obtained
concerning the ions present in certain discharges~,and some understanding of the re-
actions occurring is beginning to develop.
' In the glow discharge, the regions most studied have been the negative glow and
the positive column. Both are regions of nearly equal positive and negative ion
concentration, although the ion concentration in the negative glow is usually greater
(often 10-100 times) than that in the positive column. Further, the electric field
in the negative glow is considerably greater than that of the positive column, through
which the ions drift with relatively small energies.
Discharges in the rare gases, of course, always contain atomic ions and, at
sufficiently high pressures, diatomic ions as well. This latter can be produced
by two possible reactions, typified by helium:
*
He + He --f He: + e (6)
+ He + 2He + He2+ + He
(7)
Since the formation of He* is an excitation process it will occur over a relatively
small energy range with electrons of energy close to the ionization potential of He.
Thus, it would be expected to predominate in the positive column of a glow discharge.
This has been observed by Morris (80) and by Pahl (81,82).
(80) D. Morris, Proc. Phys. E. (London) s, 11 (1955)
(81)
(82)
M. Pahl and U. Weimer, z. Naturforsch-G, 753 (1958)
M. Pahl, 2. Naturforsch, &, 239 (195r
-
In order for reaction 7 to be observed relatively higQ pre5sures are required.
-0 2
The third order rate constant will probably not exceed 10 cc /m lecule second
-9
and the timt of the ion in the plasma will probably not exceed 10 seconds. Thus,
+ for He2 /He to be approximately 0.1 by reaction 7 the pressure must be approximately
5 Torr. Thus the formation of the diatomic ion by the three-body process will
decrease very rapidly with decreasing pressures and will be negligible below 0.1 Torr.
Knewstubb and Tickner (83) have studied the ions of the rare gases in both the negative
(83) P. F. Knewstubb and A. W. Tickner, J. Chem. Phys. 36, 674, 684 (1962)
=
glow and the positive column of a dc glow discharge. They find the ratio Ar +/Ar+
to be much less in the negative glow than in the positive column, and conclude that
the diatomic ion is formed principally by the three body process in the negative
glow, but that the chemi-ionization reaction predominates in the positive column..
Similar considerations apply to the other rare gases.
I
1
I
I

37
In our laboratory a microwave discharge was generated in helium by a 100 watt
Raytheon microtherm generator and sampled through a pinhole leak at the apex of a
+ .
conical probe into a quadrupole mass filter. The intensity of the He ion dropped
exponentially in the range studied from a relative intensity ,of unity at 0.1 Torr
to about .002 at 0.5 Torr.
+ .
In the same pressure range Hez increased in intensity from O.,OOZ at 0.1 Torr
to a broad maximum around 0.3 Torr and then declined rapidly at higher pressures.
+.
If the He2 1s forme g8by she three- ody process (equation 7) the rate constant would
9
have to be about 10 cc /molecule /sec. which seems excessive. Ne conclude then
that the diatomic ion is probably formed principally by the chemi-ionization process.
(equation 6).
Of perhaps greater interest are the ionic processes occurring in more complex
gases. Thus, glow discharges in hydrogen (84,85,) and in H,-D, mixtures (86) showed
..-
(84) C. J. Braesfield, Phys. e. 2, 52 (1928)
(85) 0. Luhr, J. Chem. Phys. - 3, lLib(1935)
(86) H. D. Beckey and H. Dreeskamp, 2. Naturforsch 9.; 735 (1954)
t
the formation of H3t or H - D mixtures, which increased in concentration with
pressure at the expense 01 the aiatomic ion. In an effort to interpret the behavior
of the discharge in hydrogen, Eyring, Hirschfelder and Taylor (87) considered the
(87) H. Eyring, J. 0. Hirschfelder and H. S. Taylor, 2. Chem. Phys. 4, 479 (1936)
reaction forming H3t to be:
Hz + + Hz +H: + H
They took the activation energy for the reaction to arise from the balance of cen-
trifugal force and polarization attraction acting in opposite directions. The result-
ing rate constant for the reaction is:
k = 2 TI e kl”
i.e., the same as equation 5. Subsequent studies of this reaction in a mass spec-
trometer ion source (4) have established this reaction beyond doubt, and have shown
that the reaction rate is given approximately by the above relation of Eyring, et al.
Recent studies by Ortenburger et al. (88) employing a high frequency discharge have
(88) I. B. Ortenburger, M. Hertzberg and R. A. Ogg, J. Chem. Phys. 2, 579 (1960)
shown that reaction 8 occurs under these conditions as well.
Studies of ions in the negative glow and Faraday dark space of a glow discharge
in water vapor at 0.4 Torr have been made by Knewstubb and Tickner (89). They
(89) P. F. Knewstubb and A. W. Tickner, J. X m . Phys. 8, 464 (1963)
-
fouqd the maximum ion intensity to occur in the negative glow. There was little
H 0 present, but a series of solvated protons was observed having the general composition
2 .
H (H O)n with n varying from one to five. Mass spectrometer studies have
established t?ie reaction: (1,2, 13)
+ +
H 0 + H20 + H30 + OH
2
(9)
and more recent studies have detected the more highly solvated species. (49,50) The
results of the discharge studies showed the first four water molecules to be more
strongly bonded to the proton than succeeding molecules and this has been substantiated
by the work of Kebarle (49,50) previously discussed.
-
-
(8)

36
Similar studies have been made of ions in a glow discharge in ammonia (90)
(90) P. H. Dawson and A. W. Tickner, J. Chem. Phys. 40, 3745 (1964)
w'th similar results. The negative g)ow at 0.4 Torr was found to contain the ions
+ H (NH3), with n from 1 to+5, with NH being formed in highest concentration in the
negative glow, but with H (NH ) preiominating in the Faraday dark space. The
3 4
multiply solvated proton has also been observed in the ion source of a mass spectrometer
at elevated pressure. (91, 92)
I
(91) A. M. Hogg and P. Kebarle, J. -. m. 43, 449 (1965)
-
(92) A. M. Hogg, R. M. Haynes and P. Kebarle, J. $. E m . z. 88, 28 (1966)
(93)
Knewstubb (93) also mentions the observation of ions in a glow discharge in
P. F. Knewstubb, "Mass Spectro~metry of Organic Ions," Academic Press, New
York, 1963, p. 284
methane in which the ions C H + and CH5+ predominated, and in which Some 40% of the
2 5
ions present contained three or more carbon atoms. This is quite different from
the results obtained by Munson and Field (47, 48) for studies of methane at elevated
pressurTs. They found that with quite pure methane at pressu es above about 1 Torr
f
the CH5 and C2H5 ,ions were present in the same ratio as CH4 and .I: (the precursors)
in the primary mass spectrum of methane, and that ions having more than two
carbon atoms were present in only minor proportions. This suggests that the ions
of higher mass reported by Knewstubb (93) originated either from impurities in the
methane employed or, more probably, from molecules such as acetylene or ethylene
formed by the action of the discharge on methane.
Nitrogen has been the subject of several investigations employing both mass
spectrometer ionization chambers and dis har es for the production of ions. The
F io s of greatest interest are N + and N4 . g The mass spectrometer studies have shown
P 3
to be formed by the reaction:
N3 -
N2 +* + N p + N:+N
where N +* imp ies an excited ion having an appearance potential of about 21-22 eV.
i.
(31,32,33) N4 has been found to result from the reaction:(94)
+ ,N2 + 2N2 + -+ N4 + N2
(94)
G. Junk and H. J. Svec, 2. &. 2. E. 80, 2908 (1958)
In addition, Munson et al. (32) showed that under certain conditions N4+ is formed
by the chemi-ionization reaction:
* +
N2 + N2 -+ N4 + e
(12)
Both ions have been obs rved in electric discharges in nitrogen. Luhr (95) and
-f
Dreeskamp (96) found N3 , but it appeared to be formed only in the drift space
(95)
(96)
-
0. Luhr, w. E. 44, 459 (1933)
H. Dreeskamp, 2. Naturforsch G, 876 (1958)
-
following the discharge. It was thought to result from the reaction
+ N + 2N2 + + N3 + N2
(13)
formed in a glow discharge at 0.3 Torr in nitrogen
+
Shahin (97) has reported N3
(97) M. M. Shahin, "Ion-Molecule Reactions in Gases," Advances in Chemistry Series
No. 58, American Chemical Society, Washington, D. C.. 1966 p. 315
-
-

+ .
4 . r
39
as well as N in a cor0 a discharge in a mixture of nitrogen and water vapor. The
relative intensity of Nq passed through a maximum at about 10 Torr, then slowly
declined at higher pressures. Shahin attributed the formation of the ion to the
\ reaction:
>
N3 + + N3 + N C
'$ In the same system he observed H 0 , H 0 , H 0; fnd N H+ ions, all apparently
2
rresulting, at least in part, from reac2ion 02 N* wit6 water.
In this laboratory nitrogen has been passed through a microwave discharge at
pressures of 0.01 to 0.3 Torr and the plasma saypled into a quadrupole mass filter
where the io s were separated and analyzed. was not observed at fny condition
studied . P
N4
N3
was formed in very small concentrations (about 1% of N2,) at 0.01
Torr, but increased rapidly in intensity, passing through a broad maximum at about
0.15 - 0.20 Torr, and the decreasing at higher pressures. In the same pressu e
f
range the intensity of N: dropped precipitously from 90 to 1.5+ and that of N
remained constant at about 9, all in arbitrary units. At the N3 maximum all of the
ions were of nearly equal intensity. A brief study of the variation of the intensi
ies of these ions with nominal power input at a pressure of 0.15 Torr showed
4 + and N to decrease in intensity with decreasing power until the discharge was
N2 .
extinguished at about 40% of maximum. These curves were reminiscent of an ionizetion
efficiency curve, and strongly fuggest the average electron energy decreased
with decreasing power input.
power input up to a broad maximum between 75 and 65% of maximum power, after which
it again decreased. This suggests that an excited state is formed by electron
impact with elec~trons of broad energy spread. Such an energy spread has been found
in our studies, and will be reported separately. Further, we find that the electrons
possess an approximately Maxwell-Baltzmann distribution of energy, and that the
average energy decreases as the pressure increases, in accordance with our observa-
+ tion of the va iation of N3 with pressure. This also accounts in part for the
f
decrease in N2 with pressure.
+
(14)
The N3 ion increased in intensity with decreasing
The near c nstancy of N with increasing pressure is difficult to understand.
?
The number of N ions formed by direct electron impact up0 N or N must be relatively
small and can hardly account for+the intensity of the N' ogserved. It is possible
that at the higher pressures most N is formed by the reaction:
N;+N + N + + N2
(15)
but this is, of course, speculative.
+
The absence of N4 is puzzling, since it has been observed in some discharges.
It is possible the ion is d stroyed in passing through the sampling probe, This seem
8 unlikely, however, since N4 is held together by a bond of about 1.5 eV energy.
This is about the same strength as the bond in He:, which we observe. Conceivably,
the ion may appear at higher pressures, but some alterations in our sampling probe
will be necessary to achieve this.
+
Seve a1 investigations of ions in glow discharges in oxygen have been reported.
0 and 0: were reported by Knewstubb (93), Luk (98) and Dickenson and Sayers (99).
L
(98) Luhr, O., m. g. 38, 1730 (1931)
(99) PT-H. G. Dickenson andJ. Sayers, Proc. Phys. =. (London) G, 137 (1960)
~~
+
In addition, 03+ and O4 (99) have been reported.
4
D
In OUK stud es of orygen In the pressure range 0.01-0.3 Torr, using a microwave
discharge only 0 and O2 were found. Both ions dropped in intensity exponentially +
with increasing pressure, but the 0' dropped slightly less rapidly than did O2
t

+
The drop in 0
with 0 by charge exchange.
2
40
with increasing pressure may be due to the fact that 0' can react
+ +
0 +02 -4 o2 + o
This would account for the drop in 0' intensity at conditions at which N+ (which
cannot un ergo loss by charge exchange) remains constant. In the same pressure
9
range, O2 drops in intensity to about the same extent as does N +, which can
disappear by charge exchange with N. 2(
In a mass spectrometer ion chamber 0 + and 04+ are found in rather smell in-
3
tensities, apparently formed by the reactions:
--f +* +
0; 2Tu +02 c o4 -3 o3 + o
+*
04 + o2 + 04+ + o2
The latter is very faint, however. The ionization potential of 0 is slightly
greater than that of O2 (100,101), so O3 can react with 0 as fo?lows:
2
o3 + + o2 + 0,' + o3
(19)
(100) J. T. Herron and H. I. Schiff, 2. Chem. Phys. 24, 1266 (1956)
-
(101) R. K. Curran, 2. Chem. Phys. 35, 1849 (1961) -
+
No doubt this accounts for the failure to observe O3 .
~ ~~
~ -~
The absence of 0; may be
merely a matter of sensitivity. t
In our laboratory we have also studied the ions formed in a microwave discharge
in a mixture of nitrogen and oxygen at a constart pressure of about 0.1 To r,
AS+
might be expected, the intensities of Npf and N decreased and those of 0: and 0
kngreased as the proportion of nitrogen decreased and that of oxygen increased.
NO was very intense ovfr th$ range $f 10% to 75% oxygen in the m xture.+ This is
not surprising, since N , 4
N2 and N, reqcting with 0 or 0 and 0 and 0 reacting
with N or N are capable of producing NO .
and no doubg is ionized by electron impact.
(16)
Further, &O is produced in ghe discharge
+
Small amounts o f NO2 ions were observed in all of the mixtures studied, and
small amounts of N 0 were found in the nitrogen rich mixtures, but disappeared
2
when the proportion of nitrogen in the mixture dropped below 75%. The manner of
their formation is not known, but from their very small intensity we infer that they
are probably formed by third order processes. It is surprising that no N: ion was
observed when oxygen was present. Presumably it is capable of reacting in several
ways with 0 or 0, and so is destroyed as fast as it is formed. The system OZ-N2
2
is of great interest and will be studied further.
This research was supported by Project SQUID of the Navy, under Grant NOM
3623 S-21, which we acknowledge with gratitude.
1
,
J
I
\I
i
\I

ION-MOIECULE REACTION RATES MEASUFIED IN A DISCHARGE AFERGLOW
E. E. Ferguson, F. C. Fehsenfeld, and A. L. Schmeltekopf
Institute for 'Peleccanmunication Sciences and Aeronomy
Environmental Science Services Administration Boulder., ' Colorado
ABSTRACT
The application of a flowing afterglow reaction technique to the measurement of
thermal energy ion-molecule reactions is briefly described. The flowing afterglow
system allows the measurement of reactions of ions with such unstable neutrals as 0,
H, N, and 03. The reaction O+ + CQ - Od+ + CO appears to be important in Cod dis-
charges, as Q+ has been found to be a dominant ion in this case.
The Of, c+, and
CO+ ions rapidly react to produce either Q+ or COa'$ the dominant ions observed in
a COz discharge. In an argon-hydrogen discharge, lb would be an important ion, ,
since it i s the most stable ion in that system and a reaction sequence leading to Ib.'
production is fast. Several associative-detachment reactions such as 0- + 0 - 02 +e
and H- + H - Hz + e have been found to have large rate constants and such reactions
may be important in determining the negative ion concentrations in discharges.
I. Introduction and Experimental
A flowing afterglow system has been utilized for the past several years in the
ESSA Laboratories in Boulder, Colorado, for the measurement of reaction rate constants
at 300' K for bot2 ositive and negative ions reacting with both stable and
unstable neutral species"'. Figure 1 illustrates one version of the flowing afterglow
tube which has been utilized. A tube of about 1 m length and 8 cm diameter
serves as the reaction vessel. A gas, usually helium, is introduced at one end of
the tube and exhausted at the other end at a rate of around 100 atm cc/sec, the
helium pressure being typically - 0.3 torr. The helium is ionized by a pulsed dc
discharge producing about 10'" He' ions and He(2S) metastable atoms per cc. Positive
ions are produced in most cases by adding a relatively smll concentration of
neutral gas into the helium afterglow by means of a small nozzle. The positive ions
are products of He+ and He(2-S) reactions with the added neutral. The reactions of
these ions with a second neutral added at a second downstream nozzle are then
measured. The ion composition of the afterglow is monitored by means of a frequency
scanned quadrupole mass spectrometer covering the mass range 1 - 100 mu. The rate
of disappearance of a reactant ion with neutral reactant addition leads directly to
a reaction rate constant.' Cur estimate of the reliability of the rate constants so
determined is f 30$ in favorable cases. Cur experience in comparing our rate constants
with other measured rate constants generally supports this estimate.
This experimental scheme has many variations. We have, for example, successfully
used Pyrex and quartz reaction tubes as well as stainless steel, and microwave
and electron beam ionization as well as the dc discharge. We sometimes find it
desirable to produce reactant ions by adding a suitable gas through the discharge
with the helium rather than downstream in the afterglow, particularly in the case of
certain negative ions such as 0- and H-, which are readily created by dissociative
attachment by fast electrons in the discharge. We sometimes use carrier gases other
than helium, particularly argon.
Some of the important features of the flowing afterglow experimental technique
are the follodng: (1) The reactant ions in many cases axe known to be in their
ground states, either because of the reaction which produces them or by virtue of
superelastic electron collisions in the plasma prior to neutral reactant addition.
Stable neutral reactant species are added without being subjected to discharge or

42
RMTS
REAC'TANT /
GAS MASS SPECTROMETER I
Figure 1. Flaring Afterglow Reaction System.
..

43
excitntion 2onditions so that they can be assumed to be neither vibrationally nor
electronically excited. On the other hand, some selective excitation is possible.
For example, the reaction
o+(~s) + N,('E) - NO+(~Z) + N( 4 s) + 1.1 eV
(1)
0
has been measured as 3 function of vibrational temperature from 300 - 5000 K.
(2) It is possible to add chemically unstable neutral reactanJs into the
afterglow so that reactions such as
I!*+ + 0 - NO+ + N, (2) 02+ + N - NO + 0,
(3)
0- + O -. o2 + e,
have been studied in this system.
(4) and 0- + o - o - + o
3 3
(3) The difficulty of resolving concurrent reactions does not arise, as it
does in mass spectrometer ion sources, and we have measured the reaction
without complication from the reaction
H2 i + C02 - C02H+ + H,
+
(5)
C02+ + H2 - C02H+ + H (6)
since H is not ionized in the flowing afterglow arrangement.
2
11. Thermal Energy Charge -Transfer Reactions
The flowing afterglow system is well suited to exothermic charge-transfer
reaction measurements, since the neutral reactant, necessarily of lower ionization
potential, does not go through the ionization region and hence is not selectively
ionized as would be the case in same experimental arrangements used for ion-molecule
rezction studies. For positive ions, either atomic or molecular, charge-transfer to
molecular neutrals is usually fast (barring occurrence of a competitive exothermic
rearrangement reaction). Very many such examples (several dozen) have been observed
and very few exceptions, notably the reaction between He+ + Hz, which is observed
not to have a rate constant as large as
cm3/ sec. For example, Ar', CO', COz',
Na+, N+, and ILOY-all charge-transfer with Oa to produce 4' with rate constants
greater than lo--- cm'/sec (or cross sections greater than 20 8").
This result contradicts most theoretical predictions which had assumed that
charge-transfer would be slow except in unusud cases inwlving fortuituous energy
resonances.
The situation appears the same for negative ion charge-transfer although much
less data is available in this case. We have found that the reactions
H- + NO2 N&- +H (8) OH- + NQ2 +
NOa- + OH (9)
0- + - 03- + 0 (10)
and several other negative ion charge-transfer reactions have rate constants greater
than lo-' cm /set n_t 300° K and previously Curran4 had observed a number of fast
negative ion charge-transfers to NOa, and Henglein and Muccini' observed a fast
negative ion charge-transfer with Sa. In the absence of experimental data one
would certainly predict that exothermic charge-transfer to a molecular neutral will
be fairly efficient.
111. Ion-Atom Interchange Reactions
The most cornonly studied ion-molecule reactions have involved changes in
molecular configuration. Typical examples are
c+ + o* - co+ + 0,

44
whose rate constant, 1.1 x 10- cm'/sec, from flowing afterglow experiments2 agrees
with an earlier value 9 x lo-' cm3/sec measured in a mass spectrometer ion SOUpCe
by Franklin and Munson"; the reaction
+ +
0 + co2 - o2 + co (12)
with a rate constant of 1.2 x lo-' cm3/sec from both flowing afterglow' and mass
spectrometer ion source measurements7 ; and
C+ + co2 - co+ + co I (13)
with a rate constant 1.9 x lo-' cm3/sec.
Such reactions are more often fast than slow. One of the slowest exothermic
ion-molecule reactions (again barring cases where charge-transfer comptes) is
o+ + N2 -t NO+ + N, (14)
with a rate constant - m3/sec3. Elis rate constant increases3 to about
3 x cm3/sec for an N2 vibrational temperature of 5000' K and also increases
with 0' kinetic energy"'.
No case of a fas? ion-atom interchange reaction involving the breaking of two
bonds has so far been reported. The exothermic reaction
02+ + N2 +
has a rate constant less than 1O-l' cm3/sec, for example.
NO+ + NO (15)
IV. Associative Detachment Reactions
A number of associative detachment reactions have been recently measured to be
fast in the flowing afterglow system (i.e. k > lo-' an3/ see), including:
and
0- + 0 + O2 + e ~
OH- + o -t H O ~ + e
0- + H2 - H20 + e
(16)
H- + H - H2 + e (17)
(18)
(20)
OH- + H - H20 + e
0- + co - C O + ~ e
Phelps and Moruzzil' have been making similar measurements in drift tube expriments
at Westinghouse and have measured reactions, (20), (21), and (22). The flaring
afterglow and drift tube results agree within better than a factor of two in each
case. Several exothermic associative detachment reactions do not occur at measurable
rates (k < 10-l' cm3/sec) , an example being
0- + N2 - N20 + e + 0.2 eV. (23) :
V. Ion-Molecule Reactions in Discharges
he obvious application of measured rate constants to the qualitative interpretation
of the ion composition of a gas discharge is the case of the COZ discharge ion
composition studied by barnon and Ticher''. Dawson and Tickner observed the d d -
nant ions in a glow discharge in COZ to be Oa+ and COZ+. This is quite reasonable in 1
xLew of the known occurrence of reactions (ll), (12) , and (13) above, together with
the fast charge-transfer of CO+ vith C& to produce COZ'.
The results of these
reactions are graphically illustrated in Fig. 2 (from Fehsenfeld, et al., J. Chem.
Rys. 5 23 (1966)), which shars that dl of the ions, C+, O+, and CO+, do convert
to 02+ and Cb+ by reaction with COa in the - 6 milliseconds reaction tbne in the
PlOWhg afterglow. If molecular oxygen were added, one would expect the dominant
ion to become Oa+ alone, since CoZ+ is horn to charge-transfer rapidly with &la.
(19)
(21)
,' '
,

45
In 2 like manner Fig. 3, showing ion composition in an Argon-& afterglod3 ,
suggests that the ion-molecule chemistry is such that the dominant ion is K3+, and
this would very likely be true for certain active discharge conditions as well.
An example of the possible importance of associative detachment reactions in
discharges was noted by Masse$', who speculated that the low negative ion density
'in oxygen dischzrges (relative to iodine discharges for example) might be due to
,electron detachment by reaction (16). In iodine the analogous reactionlI- + I -
L + e is endothermic. In view or the subsequent finding that reaction (16) is
indeed fast (kl; = 1.9 x lo-'" cm2/sec), Massey's suggestion takes on renewed interest.
The same argument could be applied to Hz discharges in view of the rapidity
i of reaction (17), kl- - lo-' cm'-/sec.
VI. Conclusions
The growing body of quantitative ion-molecule reaction rate data now available
should allow in favorable cases prediction and in many cases correlation of observed
1 ion compositions of gas discharges with known ion-molecule chemistry.
\
Ach.owledgement: This work has been supported in part by the Defense Atomic Support
Agency.
1
/
1
I
1
i
I
'>
I
,
I
\
REFERENCES
1. F. C. Fehsenfeld, P. D. Golden, A. L. Schmeltekopf, H. I. Schiff, and E. E.
Ferguson, J. Chem. Phys. 4087 , 4095 (1966).
2. F. C. Fehsenfeld, A. L. Schmeltekopf, and E. E. Ferguson, J. Chem. Phys. 44,
3022, 4537 (1966); 45, 23, 404, 1844 (1966).
3. A. L. Schneltekopf, F. C. Fehsenfeld, G. I. G i b , andE. E. Ferguson, Planet.
Space Sci. in press; J. Chem. Phys., to be published.
4. R. K. Curran, Phys. Rev. x, 910 (1962).
5. A. Henglein and G. A. Muccini, J. Chem. Phys. 15, 1426 (1959).
, 6. J. L. Franklin 2nd M. S. B. Munson, 10th Combustion Symposium, The Combustion
Institute, Pittsburgh, Pa. (1965) p. 561.
: 7. J. L. Paulson, R. L. Mosher, and F. Dale, J. Chem. Phys. 44, 3025 (1966).
8. R. F. Stebbings, B. R. Turner, and J. A. Rutherford, J. Geophys. Res. II_, 771
(1966) -
9. C. F. Giese, 152 Am. Chem. SOC. Meeting, New York, 1966.
' 10. J. L. Moruzzi and A. V. Phelps, J. Chem. Phys., in press.
11. P. H. Dawson and A. GI. Tickner, Proc. Sixth Int. Conf. on Ionization Phenomena
> in Gases, VO~. 3 p. 79 (1963) Paris.
> 12. R. B. Norton, E. E. Ferguson, F. C. Fehsenfeld, and A. L. Schmeltekopf, Planet.
Space Sci. &, 969 (1966).
, 13. F. C. Fehsenfeld, A. L. Schmeltekopf, and E. E. Ferguson, J. Chem. Phys. in
press.
' 14. H. S. W. Massey, "Negative Ions" , Second ed. , Cambridge Univ. Press, Cambridge
(1959) -

46
. ...... , .
k

47
Absorption Spectra of Transient Species
in a Single-Pulse Microwave-Discharge
by A.B.Csllear
Department of Physical Chemistry,
University of Cambridge, Lensfield Road, Cambridge.
ABSTRACT
An apparatus was described for production of an intense and
short duration microwave pulse discharge in various gases. At
power levels in the 50 kw range, the microwave energy was coupled
to the gas with a high Q tuned cavity. For higher power levels
up to 1 mw peak, a glass tube containing the gas was placed inside
a wave guide, in the central region of maximum field. Thereby
intense discharges were produced in metre long columns of gas, to
provide ideal conditions for kinetic absorption spectroscopy.
Some simple applications of the techni ue were described, including
measurement of energy transfer from Beq23s1 to atomic neon,
exchange of vibrational energy from nitric oxide to D2S, and the
production of diatomic free radicals in various gases.
Flash photolysis ie proving to be of considerable value in the
study of structure and kinetics. Our first object in developing
the microwave technique, in which the photolytic flash is replaced
by a powerful microwave pulse, was to explore alternative means of
achieving eubstantial electronic excitation of gases. It was
believed that the method would complement flash photolysis and would
also extend the direct techniques of flash spectroscopy to more
diverse systems, for example gases which only abeorb light in the
extreme vacuum ultravioletl. Thus to mention a few as yet unattained
and perhaps ambitious objects, we hope to follow dimer ormation
in pure helium and the other inert gases, to study N2 A 4 xi by
absorption spectroscopy, to investigate CH formation In CH (which
has now been achieved by vacuum ultraviolet flash photolysh *),
and to detect both positive and negative ions by absorption
spectroscopy.
A microwave field will couple only with free electrons, and
direct change of translational energy of gases is unimportant
(because of the mass restriction) compared to the energy appearing
as electronic and vibrational excitation. Described here are
results obtained with a microwave-pulse generator of peak power
-50 loa, and also the development of a more powerful apparatus with
a capability to deliver power to a gas at just under 1 m. Resulte
from the second machine will be available by the early spring of
1967.
I
$merbental an d Discussion
The first apparatus was constructed to establish the feasability
?
I

49
-
Of ProCucing transients at concentrations detectable by absorption
spectroscopy. An English Electric M561 Idagnetron (3046 Ec/sec)
was powered with a 1 JJ F capacitor charged to 15 kV, and the
pulse duration could be varied in the range 2 - 50 JJsec with
English Electric ~ ~ 2 thyratrons. 5 0
Via .a conventional 'door-knob'
Coupler and waveguide section (without an isolator), the povier Was
transmitted to a cylindrical cavity with a Q of 380. This is
illustrated by the photograph shown in figure 1, and the :main ,
details have already been describ.ed. Although a IC psec pulse
corresponds to 0,8 joules, only a minute fraction of this was,
successfully delivered to the gas. However, a number of encouraEinE
observations were made, ana some energy transfer rate coeff'icients
were measured, as described below.
Yore recently, R.E.M.Hedges, J.G.Guttridge and I have completed
the construction of a powerful microwave .generator, which .'
incorporates the English El.ectric M578 magnetron, with a rated
peak-poi;ler of 900 kw. Into a 'matched load', we have produced
single pulses at 500 kw, duration 5 bsec.,.which is close to the
maxinum power available per pulse. The H.T. arrangement is similar
to tnilt employed in the prototype equipment, with two thyratrons
delivering a potential up to 30 kV. This particular magnetron turns
out to be rather susceptible to arcing on single pulse operation,
and in this respect is especially sensitive to reflected power.
This has. necessitated the inclusion of an isolator between the
magnetron and the load; to obtain satisfactory functioning up to
25 kV (as evidenced by the profile of the current pulse), it is
necessary to condition the magnetron cathode by somewhat laboriously
pulsing with the applied potential increased by small increments
from 20 kV upwards. . .
It was anticipated that the same cavity that had been employed
in the early experiments would also be suitable for the high energy
equipment. With the assistance of hlr. F.J.!Veaver of the English
Electric Valve Comgany, the size of the hole coupling the cavity
to the waveguide was carefully optimised to give a stsnding-wave
ratio of 1.2 and a Q of about 800. With such a tuned cavity, a
discharge can be produced in heliumat 1 atmos. pressure; hoijever,
the discharge was still quite feeble and it was obvious that only
a minute fraction of the total energy was being coupled to the gas.
Although the tuning and matching was near perfect under low power
test conditions, for various reasons the cavity detunes as soon as
the gas strikes.
A41tnough a study of the cavity discharge may prove to be of
some value, interest was directed at the problem of achieving iuen-e
more efficient coupling of the microwave pulse to the experimental
gas. In fact we have now achieved conditions under which it appears
that practically the entire pulse may be absorbed by the gas.
This innovation is a very simple one; a thin walled glass tube
contqining the gas is placed inside the waveguide in the region of

50
maximum fiela. In this manner, an intense discharge can be
proaucea in metre long columns of gas, to provide ideal conditions
for absorption spectroscopy. R.E.?d.Sedges is presently
assembling the optical components to carry out spectroscopic Studies
under these conditions. The beauty of this discovery lies in its
inherent simplicity; power is coupled to the gas at low Q and
matching problems are virtually eliminated.
Results
This section relates only to the low power equipment, with a
tuned cavity discharge. Figure 2(a) shows the formation of
He(23S ) in a single pulse of IO Msec duration, and its decay
followhg the pulse. Also bserved in pulsed helinm were the
conparatively short lived 2 3 P and 2IS states. A simple quantita- ‘
tive ap lication shown in figbe 2(b),’is the enhanced rate of decay
of He(2 3 S ) in the presence of a trace of neon,
By means of
photometr3 of plates similar to that illustrated in figure 2, the
rate of the energy transfer process
He(23S1) + Ne(2’So) -> He(llSo) + Ne(2s)
2 2
reported
by Javan, Bennett and-Herriott 3. The helium system is perhaps
the simplest of all electric discharges and, as mentioned above, we
hope to make more detailed observations of the formation of diatomic
helium. All four of the Is metaables were observed in pulsed neon lo
was recorded as 0.35 + 0.02 2 , in agreement with 0.37 2
z’igure 3 illustrates vibrational excitation of nitric oxide,
and its decay following the pulseo It was concluded that excitation
occurs by direct collision of nitric oxide molecules with electrons,
because of the ex reme weakness of fluorescence from electronically
excited molecules .t The total yield of vibrationally excited NO
produced with a microwave pulse, is about 10 - I00 fold higher than
the yield of electronically excited species or free radicals, in
all the systems which have thus far been investigated. The technique
may therefore prove to be generally ap licable to the study of
vibrational energy transfer.
Figure 3fb) shows the catalysis of the
NO relaxation, by a trace of D2S, which has a vibrational frequency
close to that of NO, A number of rate coefficients for V-V
processes of this type have recently been published,
Finally we mention the formation of chemical intermediates,
produced by microwave pulses in polyatomic gaseso Thus far, these
aspects of the chemistry of electric discharges have only been I
investigated in CS (CN) H 0 and H S. In each case, diatomic
free radical inter6;diateg’wege obseraed. Some of these are
included in figure 4.
I
.f

Before
n. d.
IO.~/AS~
22.4
31.6
42.0
52.8
66.0
78, I
90-9
113
I33
I64
206
243
312
'366
Before
n, d,
22.&sec
31-6
42-0
52.8
66.0
78, I
90-9
113
(a) Formation and decay of He (2 3S,) In 5 mm .of He,
'(b) Deczy of He (2 34) In 5 mm He + I.? x IOe2,, Ne.
Figure 2

52
3
%
%-
%-
5-
a-
7
h
W
P
H
0
L
v)
a
P
o w
L
0
+
E!
c
n
i3
W
al
=:
.-
E
3
(2
rr\
+
0
b=l N
E
e
XJ
3
0
b
+
3
z
a
E
rn
n
P
U

'3 (A'ncX'Z')
53
OH (A2Zf + X2n)
>'is+ Formation ana decay of (a) cs : 2mm C S ~
+ 60 mm He;
(b) OH : 5 mm %O + 35 mm HE; (C) SH : 20 mm H28 + 100 mm He.
0.9

References I
I . A.B.Callear, J.A.Green and G.J.Williams, Trans.Faraday Soc.,
2.
3.
54
I
(1965), 61, 1831. f
W.Braun, K.H.Welge and A.M.Bass, J.Chem.Phys., (1966), a, 2650.
A.Javan, W.R.Bennet and D.R.Herriott,
6, 106.
Phys.Rev.Let.

55
The Collection of Positive Ions and Electrons by a
I , Screened Probe in the Neon Negative Glow*
I M.J. Vasilet and R.F. Pottie,
\
i
Division of Applied Chemistry,
National Research Council,
Ottawa, Canada.
i I
I INTRODUCTION
3' The collection of ions and electrons by small probes placed
iithin the plasma of a gas discharge has received considerable
ttention over the past four decades. A major advance in measurement
?echnique was reported by Boyd' in 1950 with the introduction of a
'mall screened flat probe.
The new technique made it possible to
.eparate the collected currents into the ion and electron components.
?s a result it became possible to measure their concentrations
,eparately and to extend the range of measured electron velocities to
Ieyond the ionization potential of the discharged gas with no inter-
'erence from positive ion current. The method was extended by Boyd
\nd co-workers2 and by Pringle and Farvis3 to the measurement of ions
nd electrons in both the positive column and negative glow in various
!as,, . Despite the apparent success of these studies, little or no
Ise has been made of this technique by other workers in the field.
I
i Previous work in this lab~ratory~,~ had established the
isefulness of a mass spectrometer to obtain relative ion concentrations
in the various regions of glow discharges. The work reported here is
in extension of these earlier studies and represents an attempt to
la) measure the absolute concentrations of ions and electrons in
conjunction with the mass spectrometric studies and (b) to determine
;he electron energy distributions, particularly in the negative glow
;o assist in the interpretation of processes that result in the
1.
qroduction of ions.
EXPERIMENTAL
Apparatus
f a fine wire grid that was spot welded to a 5 x 5 mm platinum frame
3
hich was mounted on a Pyrex plate. A 1/8" circular hole in the glass
~late defined the current reaching the platinum collector (.010" thick)
iounted below it. A mica insulator shielded the collector from the
ischarge. The glass plate was 1 mm thick, and the grid was a stainless
$tee1 mesh (0.001'' diameter wires) with an optical transparency of 42%.

._ . _-.. - , , . . ,. .
The Pyrex ‘discharge tube56as 50 cm long and 5.5 cm in
diameter. The electrodes were polished stainless steel discs, 5 cm
in diameter and the cathode could be moved magnetically over a
distance of 30 cm. The entire discharge tube, including side-arms, .
could be baked to greater than 35OoC. Metal bellows valves were used
for all openings into the discharge tube and the normal background
pressure of 2 x Torr was achieved with a silicone oil diffusion
PUP *
The gases were of assayed research grade and pkriodic checks
were made for impurities by mass spectrometry. The operating gas was
admitted to the discharge tube via a Granville-Phillips variable leak
valve, the pressure being measured with a Decker diaphragm gauge. A
small flow was maintained through the discharge tube during operations
by means of a needle valve to the pumps. The exit pressure was about
Torr. Preliminary discharges of about one hour duration were
always carried out before measurements were taken. The tube was then
evacuated and a fresh gas sample was admitted for the experiment.
All results reported here were taken with a gas pressure of
0.28 Torr and a discharge current of 0.250 ma.
A schematic diagram of the discharge tube and electrical t
circuit is given in Fig. 1. The potentials of the grid and collector
relative to the grounded anode could be controlled independently so
that either one could be positive or negative with respect to the anode.
The collector current passing through precision resistors (0.05%) was
measured with a 1 megohm impedance nanovoltmeter which was isolated from
ground. The minimum detectable current was 1O‘loA. The potential of
1
. the collector was measured with a vacuum tube voltmeter, while a
~ and
digital voltmeter was used for measurement of the grid potential. Grid
and-discharge currents were monitored by dc micrometers. Discharge
power’was provided by a commercial dc power supply.
B. Results
(1) Collection of low energy electrons. Fig. 2(a) is a typical semilonarithmic
plot of the electron current, showing two straight line
segments for- the thermal and secondary electrons respectively. The
space potential was obtained from extrapolation of the saturation
current to the rising portion of the curve. The gas was pure neon and
the probe was near the position of maximum electcon intensity in the
negative glow. The electron temperatures of the MaxwellIan distributions
were 0.271 and 3.67 eV for the thermal and secondary electrons, and
their concentrations were 158 x 10’ 3/m3 and 2.9 x 10’ 3/m3 respectively,
assuming collection of the random current at the space potentia16s7.
The small electron residual current that was always observed at higher
retardation potentials was subtracted from the total electron currents.
The collector potential was +7O V with respect to the anode.
2) Collection of positive ions. The collector potential was
maiAtained at -70 V and the grid voltage was varied from -70 V to Well
above the space potential for the collection of positive ions. The
method of calculation of the random positive ion current differed
somewhat from previous approaches and is given briefly below. We
assume that positive ions are accelerated from the sheath to the grid
and are further accelerated by the full potential between the COlleCtor
the grid. The current rises until all the random Ion current
arriving at the sheath edge has been collected. Further collection of
Ion current is the result of sheath expansion, edge effects and
potential penetration of the sheath. If we treat the ion currents
. .
. .
I
,‘
1
,A
1
!
I
1

I
\
\
I
i
h
FIG. 1
I
NVY
57
ANODE
I
e00
io00 .OB
IOOK 1%
IOOK in
Apparatus and schematic circuit diagram.

58
0 0 0
VI
J
>
I
cv .
u
H
a

elow the saturation limit we find59
in which n is the random ion density, q’ the electronic charge
and Vf thePvelocity of the ions arriving at the collector. This is
given by:
in which Vo is the initial ion velocity and Vacc is the potential
difference between the plasma and the collector. Thus:
Therefore, a plot of i+ versus Vacc1/2 should yield a straight line
whose slope is
+ np q(?)’’2 .
The positive ion currents, plotted in this manner always yielded a
straight line to a break point which was assumed to be the saturation
limit. A plot of this type is shown in Fig. 2 (b); the small residual
ion current has been subtracted from the experimental points. From
this slope we calculate np = 145 x 1013/m3 in good agreement with the
electron concentration of Fig. 2 (a) for the same probe position.
(3) Axial variation of ions and low energy electrons. The observed
ion and electron concentrations as a function of axial distance from
the cathode are given in Fig. 3 for the probe facing the anode. There
is a slight displacement between the maxima for the two distributions
and this may reflect a slight penetration of the field from the cathode
dark space. However the results show that a true plasma is present in
the negative glow, since to a gbod approximation ne = np.
electron current was relatively insensitive to position of the probe
and gave an average concentration of 3 ? 1 x 10 3/m3 with electron
temperatures in the range of 3.5 - 4.0 eV. The rising portion of the
secondary electron curve marks the anode edge of the negative glow. At
this point the electric field begins to accelerate thermal electrons
to the secondary electron velocities. For example at d = 7.2 cm, the
secondaries account for almost 50% of the total electron current.
(2)
(3)
The secondary
(4) Residual electron current. Small residual electron currents
were observed ,for all probe positions and orientations. These are
summarized in Fig. 4 for pure neon. These minimum currents were
ccnstant for a range of -35 to -50 V in grid potential. For grid
voltages in excess of -50 V there was a gradual rise in the current,
presumably as a result of grid emission by the bombardment of positive
ions. We find in Fig. 4 that the residual current decreases
exponentially throughout the negative glow with increasing distance
from the cathode for both orientations of the probe surface. Additionally,
there is a reduction by almost a factor of 5 for the probe
facing the anode versus cathode position.
We have also observed no
variation in the residual current as a function of radial position from
the center to a point only 0.6 cm from the wall. The current increases
slowly with axial distance in the Faraday dark space (d > 5 cm) and
reaches a plateau in the positive column.

.... .
0 2
e
I /
m
0
H
F4
:
1

62
The effect of adding O.O* xenon to neon is seen in Fig. 5.
The residual current drops and there is a steeper slope to the exponentti
part of the curve. In addition a smaller increase is observed in the
Faraday dark space.
(5) Energy distribution of secondary electrons. From Fig. 2 (a) we
see that the energy distribution of secondary electrons is essentially
Maxwellian at the indicated position of the probe. However, as the
axial distance is increased we find a successively greater loss of
electrons whose energies exceed about 10 eV, although thd distribution
is not radically altered from Maxwellian. The addition of 0.07% xenon
results in almost complete depletion of electrons above 12 eV and there \I
are considerable deficiencies beginning only 5 eV above the space
potential.
The data are summarized in Fig. 6. The distribution curves
for the mixture were calculated from the experimental measurements
according to the method of Medicus8, and comparison plots of the ideal
Maxwellian distribution are given. The two Curves in each case were
normalized to those regions along the potential axis which gave a I
straight line on the semi-logarithmic plot. The fluctuations in the
curves for the mixture were reproducible and are similar in appearance ‘I
to the structural features that were observed by Twiddyg in the cathode
region of rare gas discharges. It is apparent that xenon effectively
reduces the higher energy secondary electrons, even at 0.07% concentration.
Twiddy’O has shown that there is a similar loss of energetic
electrons in the positive column of an argon-neon mixture.
1
(6) Residual positive ion current. The ratio of residual currents
for ie m i d i + min was 5:1 with the probe facing the cathode and about
1O:l with the probe facing the anode. These ratios were constant for
all axial positions of the probe in the negative glow and in the Faraday
dark space. The ratios for the 0.07% xenon-neon mixture were the same
as for pure neon.
DISCUSSION
Since the production of ions and low energy electrons in
the negative glow is governed largely by the rate of arrival of high
energy electrons from the cathode dark space it is obvious that the
direct observation of these high energy electrons offers an important
method for interpreting the processes of ionization in the negative
glow. A possible means of making such a measurement is t o consider the
negative residual currents collected by the screened probe when it faces
the cathode. These currents could arise from several processes:
(a) Direct collection of high energy electrons.
(b) Grid emission by high energy electrons.
(c) Grid emission by bombardment of positive ions.
(d) Grid emission by metastable atom impact.
(e) Grid emission by photon impact.
Rotation of the probe to face the anode caused a 2/3 redyctio4
in the-residual current. One would not e)tpect processes (c), (d), and
(e) to depend markedly on probe orientation so that these processes 1
probably account for less than 1/3 of the total residual negative
Current when the probe faces the cathode. Since neither- the positive ‘1
ion density nor the visible photon intensity decays exponentially,
Processes (c) and (e) are similarly rejected as a major source of curred
for the anode orientation. We conclude that the major causes of the
negative residual current in the negative glow are processes (a) and (b) i
,
),
t
1

\
0.9
0.8
0.7
06
0.5
0.4
0.3
0.2
01
0.8
0.7
0.6
0.5
04
0.3
0.2
0.1
0.9
0.2 02
0.1 01
I c)
2 4 6 8 IO I2 14 . 16 18 2C
ELECTRON ENERGY I V-VSP~CE)
FIG. 6 Energy distribution function of secondary elec-
trors compared with the Naxwellian distribution.
(a) pure neon, d = 2.3 cm(b) mixture .07$
Xenon in neon, d = 2.2 cm(c) mixture, d = 3.0 cm.
Probe facing anode.

64
for the probe facing the cathode. We Cannot distinguish between (a)
and (b), nor is it possible to calculate the density of such high
energy electrons without knowledge of their velocities.
The observed exponential decrease in ie min through the
negative glow of pure neon (Fig. 4) can be written as:
id = io exp(-a n(d-do))
I
in which io is the residual current at do cm from the cathode, d is
the distance of the probe surface from do and n is the concentration
of neon atoms per unit volume. Thus a has the dimensions of cm2 and
is an experimental cross section for loss of high energy electrons, as
a first approximation. From the slope of the exponential curve in
Fig. 4 we calculate that a = 8.7 x cm2. This is a reasonable
value since the cross section for ionization of neonll by 100 eV
electrons is 7.58 x cm2.
The addition of 0.07% xenon brings about a 30% reduction in I
the negative residual current and the exponential slope increases so
that the experimental a = 1.03 x cm2. This cannot be accounted /
for by assuming that the only additional loss is that ar sing from the
ionization of xenon. At a concentration of only 7 x atom %, this
would imply a cross section for ionization of xenon equal to 228 x 10-'6 ,
cm2 - about a factor of 40 higher than the known of xenonll for 100 eV
electrons. A more reasonable inference is that the loss of neon metastables
by the Penning reaction:
Ne*(3P2,0) + Xe * Xe+ + Ne + e- (5)
is responsible for the lower values of ie min, and that the change in
slope more nearly reflects the actual stopping power of neon for high
energy electrons. Since excitation as well as ionization can result
from impact of a high energy electron, a cross section of the order of
1 x cm2 is of the expected magnitude for an electron of approx-
imately 100 eV.
It would be expected that the addition of a small amount of
xenon would not significantly change either the cathode fall or the
number of directed high energy electrons passing into the negative
glow. Thus, if one assumes that the reduction in the residual electron
current is caused by reaction (5) it is possible to obtain an approx-
imate value for the Concentration of metastables in the beginning of
the negative glow.
Referring to Fig. 4, for d = 2 cm, -Ai, the reduction in
the residual current is 0.02 VA. If the neon metastables have thermal
velocities, and if one considers a reasonable yield of about 0.03 for
the grid emission (geometric factor x efficiency), then a minimum
concentration of 1 x 10i3/m3 is obtained for the metastables. It is
then possible to calculate the reactive collision frequency between
Ne* and Xe by assuming the same cross section for deactivation as
Sholette and Muschlitz12 observed for the He* - Xe Penning reaction,
namely 12 x cm2. This results in a calculated lifetime for the
neon metastables of 0.5 m sec. This Is somewhat shorter than the
lifetime of 0.875 m sec measured by Blevis et a1.13 for the decay of
Ne(3P2) in pure neon.
We conclude that xenon can intercept the
metastable neon atoms in the required time and that our assignment of
the Penning reaction is reasonably justified as a major contributing
(4)
I
1
i(

65
' cause of the reduction in the residual negative current. Similar
argUmentS can be used to account for the.decrease in the residual
current when the probe faces the anode. The effect is even more
' marked in this case, particularly at the beginning of the positive
column.
The treatment of the positive ion residual current is not
' as straightforward, since there arises the additional complication
7 ,of electron impact ionization in the space between the grih and the
collector. To a rough approximation we can account for about 60% of
i' the current with the probe facing the cathode by this process, but
{ the remainder can be a combination of secondary emission from the
? collector by photons, by metastable atoms and by energetic electrons.
' The very small currents observed for the probe facing the anode suggest
that the release of electrons by metastable impact is much less than
9 would be expected for the metastable density obtained above. Apparently
i the impact of metastables with the platinum surface produces secondary
electrons with a much lower efficiency than does the stainless steel
1 grid. The reason is not clear, but a contributing cause may be due to
the higher work function of platinum.
\
I
1
4
1.
'~ 2.
' 3.
; 4.
~ 5.
6.
; 7.
\
1. 8 .
t 9.
REFERENCES
N.R.C. No. ...., 'N.R.C. Postdoctorate Fellow.
R.L.F. Boyd, Proc. Roy. SOC. m, 329 (1950).
R.L.F. Boyd and N.D. Twiddy, Nature 173, 633 (1954).
D.H. Pringle and W.E.J. Farvis, Proc. Phys. SOC. =, 836 (1955).
P.F. Knewstubb and A.W. Tickner, J. Chem. Phys. 36, 674 (1962).
P.F. Knewstubb and A.W. Tickner, J. Chem. Phys. 36, 684 (1962).
I. Langmuir and H. Mott-Smith, Jr., G.E. Review 2J,449 (1924).
H. Mott-Smith and I. Langmuir, Phys. Rev. a, 727
G. Medicus, J. Applied Phys. 27, 1242 (1956).
N.D. Twiddy, Proc. Roy. SOC. E, 379 (1961).
.;' 10. N.D. Twiddy, Proc. Roy. SOC. a,
338 (1963).
(1956).
i 12. W.,P. Sholette and E.E. Muschlitz, Jr., J. Chem. Phys. 36,
11. D. Rapp and P. Englander-Golden, J. Chem. Phys. '13, 1464 (1965).
3368 (1962).
I
i
13. B.C. Blevis, J.M. Anderson, and R.W. McKay, Can. J. Phys. 35,
941 (1957).

66
ATTACHMENT OF MOLECULES WITH HIGH PERMANENT
DIPOLE MOMENT TO IONS IN GASES
AND INDUCED
P. Kebarle, G.J. Collins, R.M. Haynes and J. Scarborough
Department of Chemistry, University of Alberta, Edmonton, Canada
ABSTRACT I
The mass spectrometric study of clustered ions in function of polai
gas pressure, temperature and electric field strength provides informa-
tion not only on what ionic species might be present, but also on the 1
strength of the ion-dipole interactions. Studies of the reaction
H30+.(n-1)H20 + H20 = H30+.nH20 in a Nier type ion source with a 100 ke-c
proton beam are compared with earlier work on an alpha particle mass .
spectrometer. The residence time of the ions in the two sources are a
few psec and a few msec resp. Similarity of results shows that cluster,
ing equilibrium is established within psec for water clusters above 0.5
torr. The presence of electric fields reduces the cluster size. 50 1
volts/cm (at 1 torr) strip the clusters to H 0'. Experiments on the cor
parative attachment of different molecules S~OW that the stability of tl
attachment increases in the order H20, NH3, CH30H, CH3N02 for first she!
attachment. At larger distances from the ion (second shell) water be-
comes the strongest attaching. These results are explained by differenc
of dipole moments and polarizabilities.
'I
/!
/.
I
I
,,-

INTRODUCTION
I' Attachment of polar molecules to ions in the gas phase occurs in a
mber of systems: gases under high energy irradiation, gas discharges,
IlameS, etc. The polar molecules may be present as an accidental imkrity
or as a deliberate addmixture. The attachment has a number of
$Portant effects such as change of mobility, change of rektivity with
2gard to ion-molecule reactions, change of the positive (negative) ion
tlectron) recombination coefficient and change in the nature of the
-0ducts resulting from the positive negative ion recombination. The
+dy of ion-polar molecule interactions in the gas phase can also pro-
:de information on ionic solvation since the formation of ion clusters
2nstitutes the first and most important step in the solvation of anion.
; A systematic mass spectrometric study of ion-polar molecule inter-
:tiOnS was started in this laboratory some five years ago.8-12
The mass spectrometric gas phase studies are based on measurement
! the relative concentrations of the clustered ionic species: A+.nS,
+.(n+l)S, etc. The measurement of the relative concentrations is obfined
by bleeding a probe of the gas into an ion mass analysis system,
.,e. a vacuum chamber attached to a mass spectrometer. In the vacuum
hamber .> the gas is pumped out while the ions are captured by electric
ields, accelerated and focused and then mass analysed by some convenional
means (magnetic separation quadrupole filter, etc.). After mass
qalysis, the ion beam intensities are detected as electrical currents.
Several types of solvation studies can be undertaken if the relaive
concentrations of the ionic species are known.
1. Solvation Enthalpies and Entropies of Individual Solvent Mole-
ule Additions Steps: Consider the ion A+ produced in the gas phase by
3me form of ionizing radiation or thermal means. If the atmosphere
urrounding the ion contains the vapor of a polar molecule (solvent S),
number of clustering reactions will occur.
i
A+ + S + A+.S (0,1)
I A+.S + S + A+.2S (1,2)
,~
~ where
+
A+.(n-l)S + S + A+.nS (n-l,n)
At equilibrium the following relations will hold
AFo = AFo + AFo 1,2 ... + AFon-l,n
O,n 0,1
(1)
AFon-l , n
= -RT In
'A+. nS
'A+. (n-1) s-'s
= -RT In Kn-l,n (11)
P, is the partial pressure of X.
Thus knowledge of the equilibrium concentrations of the clustered
pecies A+.nS obtained from experiments at different pressures of s w ill
llow the determination Of Xn-1 n and AFn-l,n. Such measurements done
k different temperatures will lead to the evaluation of AHn-l
and
Sn-lIn. The availability of such detailed information will, #or In-
,tance, reveal the shell structure since a discontinuous change of the
Hn-1 n and ASon-l,n values will occur whenever a shell is completed.
,inally, the total heat of solvation of the ion can also be obtained
rom the expression 111, with equations of the same form holding for the
/ m

free energy and entropy change of solvation.
It is evident from equation I1 that only the relative concentrations
of the ionic species are required. Thus AFon,l and Kp can be obtained
from equation 11 by assuming that the mass
'
spechometrically measured
ion intensities are proportional to the equilibrium parti 1 pressures of
the ions in the ion source. The solvation of NHa by NH3 and H3O+ by
H20 l2 represent examples of studies of this type.
I /
2. Comparative Solvation of Two Different Ions by the Same Solvent:
There are three variants of this type of study. 11
+
Comparative solvation of ions A , C+ by solvent S.
(a)
The two ions are produced in the same system which also contains
vapor of the solvent S. In general, depending on the effective radius
and structure of the ion, the relative concentration of clusters A+.ns
will be different from that of C+.rnS. Thus, for example, in the average
n may be larger by 1 or 2 units than m. This will reveal a stronger interaction
of A+ with S. Comparison of n and m can be done at different
temperatures and pressures (of S) so as to compare the interactions in
the inner shell or in the outer shells. An example of this type of study
is the system ~30+, NHa fyd Na+ in H20 vapor which is described in a
forthcoming publication.
(b)
Comparative solvation of ions A+ and B- by solvent S.
An example of this type of study would be the system K+ and C1'
with H20. Since the orientation of the water dipoles,is reversed in the
solvation of positive and negative ions, such a comparative study is of
great interest, particularly for isoelectronic pairs a8 the one quoted
above. We have not yet done studies on such isoelectronic pairs. However,
such studies are perfectly possible. The pair K+ and C1' could be
produced in water vapor. By reversing the mass spectrometer controls,
measurements on the positive and negative ions in the system could be
done within minutes of each other.
(c) Comparative solvation of two negative ions by S.
An example of this type of study for the ions Cl', BCl-, and B2C1'
by H20 is given in a forthcoming p~b1ication.l~
3. Competitive Solvation of Ion A+ (or B-) by Solvent Molecules of
Solvents S and S : A comparison of the solvating power of two different
solvents cb-tained by measuring the composition of ion clusters
when two different solvents are present at known partial pressur 8. An
example of this type of study is the competiti e solvation of NIig by HqO
and NH3 which has been reported on elsewhere.2C Some more recent studies
will be described in the present paper.
These concern the competitive
solvation of the proton by water and methanol. The water methanol system
is of interest since the dipole moment of water is slightly higher than
that of methanol while the polarizability of methanol is considerably
larger than that of water. Thus it might be expected that at close range
to the central ion methanol might be preferentially solvating. I
' ,
1
1
'
I
1

I
69
EXPERIMENTAL
The mass spectrometric study of ion solvent molecule interactions
requires mass spectrometric apparatus which can sample ions originating
in ion sources at relatively high pressures. Three somewhat different
arrangements are presently in use in this laboratory.
1. Alpha particle mass spectrometer
11
A recent version of this apparatus has been described elsewherel’
in detail. The gas, supplied from a conventional gas handling system,
is irradiated in the ionization chamber. The radiation is supplied
from an enclosed 200 mc polonium alpha source. The irradiated gas bleeds
through a leak into the evacuated electrode chamber. There the ions
carried by the gas are captured by the electric fields while the gas is
pumped away. The ions are focused, accelerated and then subjected to
mass analysis and electron multiplier detection in a 90° sector field
analyzer tube.
2. Electron beam mass spectrometer
6
In a more recent apparatus an electron beam is used as ionizing
medium. The ion source is identical to that used with the alpha source
with the exception that th former alpha source port contains only one
very thin nickel foil (lo-‘ inches) window through which the electrons
enter the source. The electrons are created by an ordinary electron
gun housed in a sidearm of the vacuum chamber opposite the nickel window.
The electron filament is kept at some -25000 volts while the ion
source is near ground potential. Absence of radioactive contamination,
high intensity (- 10 microamps) and possibilities for pulsing (for determination
of ionic lifetimes) are some of the advantages of the electron
beam source. A disadvantage is the much greater scattering of
electrons at high ion source pressures which makes beam collimation a
problem. A quadrupole mass analyzer is used with this instrument.
5
3. Proton beam mass spectrometer
A 100 kev proton beam obtained from a Walton-Cockroft accelerator
is used as ionizing medium. The ion source is not like the previous ones
but of the conventional design, i.e. a rectangular box with repeller
and narrowed-down ion exit slit. The proton beam enters and exits the
ion source through very thin nickel foil windows (10-5 inches).
ion optics are of the conventional Nier type and magnetic analysis is
used. A proton beam (preferably of even higher energy than used by us
since at energies below 100 kev charge exchange converts appreciable
fractions of the proton beam into an H atom beam) seems to be the most
convenient ionizing medium for high ion source pressures since it pro-
vides high intensity, possibility for pulsing, and little scattering at
high pressure. The proton beam can also be deflected electrostatically
before entering the ion source. This permits variation of the proton
beam - exit slit distance in the ion source. The cost of Walton-
Cockroft accelerators is relatively low.
A high pressure mass spectrometer using Mev rotons from a Van de
Graaff accelerator has been described by Wexler. 13
The
? -

70
RESULTS AND DISCUSSION
+
1. Competitive solvation of CH3E2 by water and methanol I
A study of the ions in methanol vapor in the pressure ran e 1-10 t
torr showed that most of the ions belonged to the series CH30Hq.nCH30H.
This sug ested the possibility of observing the competitive solvation
of CH30H$ by water and methanol molecules. Figure 2 shoys the relative r
intensities of the mixed clusters obtained when irradiating mixtures of
5% methanol: 95% water and 20% methanol: 80% water, both at 5 torr totall
pressure and 5OoC ion source temperature. In the mixed methanol-water
clusters the question of the structural assignment arises. Thus the
ion of mass 119 could be assigned as H+.2CH30H.3H20 or CH OH3.CH30H.3H20,
or H30+.2CH30H. 2H20. We have selected the notation CH30Hi.CH30H. 3H20
(or CH30H3.mM.wW, for the general cluster where M and W stand for CH30H ,
and H20 and m and w for the number of methanol and water molecules). [,
The proton was assigned to the methanol oxy en ion since the proton af- ,
finity of methanol is some 10-20 kcal/mole q3 higher than that of water
This assignment is also consistent with the observation that even in
mixtures where water predominates, i.e. 5% methanol, the pure h drate '
H30+.wW was not observed, while the pure methanol cluster CH30H2.m
was observed.
The clusters obtained with 5% methanol (Figure 1) contain, on the
average, considerably more methanol than water even though the partial
pressure ratio of water to methanol is 19:l. Thus methanol is the
stronger solvent in the observed clusters, i.e. clusters containing up
to six solvent molecules. We shall be able to understand the meaning
of this result better after a more detailed treatment of the data. It
can be shown that the distribution of water and methanol in the observed
clusters follows quite closely a probability distribution. Calling the
probabilities for inclusion of water and methanol w and p, for a cluster
with a total of II solvent molecules the probability distribution will
be given by the binomial expansion of the term (w+p) E. For example if
a probability distribution is followed, the cluster containing three
solvating molecules CH30HZ.mM.wWl where II = m + w = 3, should show the
following relative intensities: sH30H5.3W : CH30H;. 2WM : CH3OH5.WZM :
CH30H2.3M = m3 : 3~ p : 3wp2 : p . We have obtained values for u and
by fitting binomial expansions to the experimentally observed distri-
bution. The calculated intensities shown in Figure 2 demonstrate that
a relatively good fit of the experimental data can be obtained. In
order to express the preference for inclusion of methanol and water per
unit methanol and water pressure we define y = (p/p~)/(w/P~) as the
factor for preferential take up of methanol, PM and Pw being the p a r d
pressures of methanol and water present in the ion source. The y's tal-
culated in this manner are given in Figure l. The results at 5 and 20%
methanol show that the yls for a cluster of a fixed size (i.e. e = COnt)
are approximately independent of the methanol-water pressure ratio.
This independence was confirmed in a number of other runs with 2, 4, 5,
8, 20, 50% methanol at 2 and 5 torr total pressure. The y's are found
to decrease as e increases. Thus methanol is taken up preferentially
by a factor of 55, 21 and 8 for clusters containing 3, 4 and 5 solvent
molecules. Figure 2 shows a log y plot versus the number of solvating
molecules. The plot is almost linear and allows an extrapolation to
log y = 0 or y = 1. This occurs when the cluster contains seven Sol-
vating molecules. For e > 7, y becomes less than unity, i.e. water be-
gins to take precedence. Figure 2 also shows results obtained with the
proton beam mass spectrometer. The total pressure in these runs was
much lower (0.6 torr) and the reaction time much shorter (a few micro- ,
seconds versus a few milliseconds in the alpha particle ion source).
'5
.!

+
Figure 1 Mixed water and methanol content in CH30H2.mCH30H.wHZ0
clusters. Ion source temperature 5OOC.
2 .o
1.5
0.5
0
2 3 4 5 6 7
n= Number of Solvating Molecules 1
Figure 2 Plot of log y versus number of solvating molecules.
y is factor for preferential take up of methanol over
+
water into ion CH30H2.mCH30H.wH20, where m + w - tl.

72
One might suspect that under these conditions clustering equilibrium
might not be achieved. However the results are, on the whole, quite
1
similar to those obtained with the alpha ion source. This might be
1
taken to mean that clustering equilibrium establishes very rapidly and
that the proton beam results approach equilibrium.
In the interpretation of the present results considerable help can
be obtained from the "electrostatic theory" for metal-ion coordination
complexes.2 It is recalled that this theory, using simdle electrostatic '
concepts, allows the calculation of the binding energies of metal complexes
in the gas phase and that the results of such calculations have
been in many cases very successful. In general, the potential energy
I
of a complex ion is built up of four terms. These are due to the attrac- ''
tion between the ion and the permanent and induced dipole of the ligands,
the mutual repulsion of the dipoles, the energy required to form the
induced dipoles and the van der Waals repulsions between the ligands
and the central ion. Comparing the potential energies of an ion having
water or methanol molecules as ligands, it is found that the first term
is the decisive one. The first term contains the sum of the permanent
dipole and the polarizability. The dipole moments of water and met anol
are 1.85 and 1.69 D while the polarizabilities are 1.48 and 3.23 b. 9
The potential energy of an ion dipole interaction varies inversely with
the square of the distance while the polarizability interaction depends
on the fourth power. It follows that the methanol molecules with their
slightly lower dipole but considerably higher polarizability will be
more strongly solvating than water at close range to the ion. The experimentally
observed preference for methanol is thus to be understood
as resulting from the higher methanol polarizability.
It can be shown that the possibility of fitting the observed clusters
with a given II by a probability distribution suggests that for
!? < 6 all solvent molecules belong to an inner solvation shell. Suppose
that for !? = 5 some of the molecules went into an inner shell (fully
occupied) and the rest into an outer shell. The preference for methanol
over water in the inner shell will be very different from that in the
outer shell. It might even be expected that water will be preferentially
taken up in the outer shell. It follows that the water-methanol
distribution in a cluster containing inner and outer shell molecules
could npt be fitted by a single probability distribution of the type
(W + p) but that the inner and outer shell would have to be fitted
separately. Such a case is in fact observed in the Eqmpetitive sOlVation
of NH$ by H20 and NH3 which was studied earler. Since the watermethanol
clusters of !? = 3, 4, 5 can each be fitted with a probability
distribution we can conclude that these clusters do not contain molecules
which occupy a distinct outer shell, i.e. that the inner shell
' contains at least five solvent molecules.
The ability to fit a cluster of constant k with a probability distribution
is, to a certain extent, surprising even if all molecules
belong to the same solvation shell. A probability distribution means,
for example, that in the five cluster the preference for methanol over
water is the same whether all the remaining four ligands are water Or
methanol or a mixture of them. Obviously this can not be strictly true-
The meaning of the experimental result must be that the nature of the
other occupants is , in the first approximation, not important.
While y remains approximately constant for a cluster distribution
with k = const, it was observed that Yp=COnst decreases from II = 3 to
J
k = 5. This can be understood if one assumes that whenever II is increased
by one unit the effective radius of the (inner) shell increases. This
causes the polarizability to become less important and leads to a decrease
of the preference for methanol. An increase of the effective
I
i
14
i
I

73
' , radius might be expected because of the mutual repulsion due to dipole
and van der Wall's forces between the ligands.
i Experiments on the competitive solvation of CH3NO2 and H20 have
also been made but this work is not yet completed. The available results
show that CH3NO2 is taken up preferentially at close range of the central
( ion (NH; and H3O+ gave similar results). Thus at room temperature the
\ preference factor is - 200 for three ligands, 60 for four Ligands, 40
for five, etc. The results also seemed to show that nitromethane is
always taken up preferentially in an incomplete.shel1 (inner or outer)
but that it is ultimately expelled from inner shells.
1
,
2. Effect of electric fields on cluster size
One may expect that the presence of electric fields will lead to a
decrease of the average cluster size. This effect is illustrated in
Figure 3. The data were obtained with the proton beam mass spectrometer
which has a Nier type ion source, i.e. ion exit slit and repeller. The
distance from proton beam to ion exit slit was II 0.4 cm. E, is the repeller
field strength. In preliminary experiments at very low repeller
fields it could be shown that at H20 pressures higher than about 0.3
torr the cluster composition obtained with the proton mass spectrometer
was very similar to that obtained with the alpha particle instrument.
Since the estimated ion residence times are - 10 micron for the proton
mass spectrometer and - 1 millisec for the alpha instrument these results
were taken to mean that in both instruments clustering equilibrium
or near equilibrium is achieved. Figure 3 shows that increase of the
electric field decreases the cluster size. This is a gradual process
as shown by the succession of maxima. The maximum composition shifting
successively from four water to zero water content of the H~O+ ion.
Possibly plots of the type given in Figure 3 could be used for determination
of the enthalpy of attachment, i.e. AHn-l,n.
-f LITERATURE CITED
I
1. A.P. Altshuller, J. Am. Chem. SOC. 77, 3480 (1955).
2. F. Basolo and R.G. Pearson, Mechanisms of Inorganic Reactions, p.64
(John Wiley & Sons, New York, 1958).
' 3. R.P. Bell, The Proton in Chemistry (Cornel1 University Press, Ithaca,
t New York, 1959).
z
4. M. Born, 2. Physik, 1, 45 (1920).
5. G.J. Collins and P. Febarle, to be published.
6. D.A. Durden and P. Kebarle, to be published.
7. W.J. Hamer (Edit.) "The Structure of Electrolytic Solutions" Chapter
5 (John Wiley & Sons, New York, 1959).
8. A.M. Hogg and P. Kebarle, J. Chem. Phys. 43, 449 (1965).
9. A.M. Hogg, R.M. Haynes and P. Kebarle, J.X. Chem. SOC. 80, 28 (1965).
, 10. P. Kebarle and E.W. Godbole, J. Chem. Phys. 39, 1131 (196n.
,11. P. Kebarle, R.M. Haynes and S.K. Searles, Advances in Chemistry, 58,
, 210 (1966).
\12. P. Kebarle and A.M. Hogg, J. Chem. Phys. 42, 798 (1965).
13. P. Kebarle, Advances in Chemistry: Applications of Mass Spectrometry
to Inorganic Chemistry (to be published) (American Chemical Society).
)14. M.S.B. Munson, J. Am. Chem. SOC. 87, 2332 (1965).
15. S. Wexler, Advances in Chemistry,TL, 193 (1966).
I
I
J
1-

z
0 c
a
60
2 40
0
2
U
+
2
5 20
M
0
Figure 3
74
036 torr HzO
0 10 20 30 40 50
ION ENPRGY, E,I (volts)
Effect of electric field on cluster composition.
Ere equals voltage between repeller and ion source exit slit.
Temperature of ion source 3OOC.
Water pressure 0.36 torr.
Experiments taken with 100 kev proton beam mass spectrometer.
!
I
II

75
Negative Ion Mass Spectra of Some Boron Hydrides
D. F. Munro, J. E. Ahnell, and W. S. Koski:
Department of Chemistry, The Johns Hopkins University, Baltimore, Maryland 21218
Abstract I ,
Questions of the role of negative boron hydride ions frequently arise in
problems concerning energetics of electron impact processes, discharge tube reactions,
and radiation chemistry. In view of the fact that the boron hydrides are
electron deficient compounds, it might be expected that they are more prone to
negative ion formation than hydrocarbons; however, very few studies of such ions
have appeared in the literature.
This paper gives some qualitative results obtained
on boron hydride negative ions produced by electron bombardment of diborane, tetraborane,
and pentaborane-9. A pulsed electron beam, a quadrupole mass analyzer, and
an electron multiplier were used for this work. In diborane, the most inteise ions
observed were BzH3- pd BHQ-.
B*H-, Bz-, BH3-, BH2 , BH-, and B were also readily detectable. Similarly, the -
negative ions observed in tetraborane - included the relatively high intensity B3H7
and lower intensities of BzHg-, B2H4 , BHI, ,,etc. In addition, indications were
obtained for B4Hg . Similar results were obtained with pentaborane-9.
Lover and varying intensities of BzH4-, B2H3 , B2H2 ,
Introduction
For many years, mass spectroscopists have dealt almost exclusively with processes
involving positive ions; ionization, fragmentation, and ion-molecule reactions.
Recently, however, despite problems of low intensity, an increasing amount
of work has been done on negative ions. A number of species have been character-
ized,' and a few reactions have been studied.2s3
In addition, at least one investi-
gation on the role of negative ions in flames has been carried out, using mass
spectroscopic techniques.
The boron hydrides are traditionally classified among the electron deficient
compounds, and might well be expected to form negative ions. whennore, some work
has been done with the boron hydrides in gaseous discharges and in radiationchemistry
where these ions might conceivably play a role.5
Thus, it was of interest to us to
study the negative ions of the boron hyhrides with the eventual hope of providing
information to permit the evaluation of the role these ions might play in discharges
and other associated phenomena.
We have under construction in our laboratory a tandem mass spectrometer for the
study of ion-molecule reactions. In the course of testing the first stage of the
instrument, it became apparent that the spectrometer might be suited for the observation
of negative ions. The ions from SF6 and 02 were detected with relative ease.
It was decided to investigate vhat negative ions could be observed in the simpler
boron hydrides; diborane, tetraborane, and pentaborane.
Because the spectrometer is still in the construction stage, some of the re-
finements normally found on conventional instnunents are lacking, so that the
results obtained are necessarily of a qualitative nature. Nevertheless, it was felt
that the observations were sufficiently interesting to warrant reporting.
Experimental
The important features of the basic experimental arrangement are shown diagremmatically
in Fig. 1. The gas to be analyzed flows from a bulb at roan temperature
into the ion source through a Granville-Phillips leak. The ion source, which is
similar in design to one described by Von Zahn,6 has an electron beam energy

76
variable from 0 - 100 volts. There are no provisions for measuring pressure,
temperature, or ionizing current in the ion source. The ions are focused and
accelerated to 50 - 100 volts, and mass analyzed by a quadrupole mass filter with a
length of 10.0 inches and an ro of 0.27 inches. Mass scanning is accomplished by
varying the RF and DC voltages to the quadrupole. The resolution of the quadrupole
is electronically variable. The ions are then accelerated to 600 volts and detected
by a Bendix Channeltron continuous electron multiplier.
Because of the absence of slits in the mass spectrometer, the lrelatively open
nature of the ion source, and the high sensitivity of the detector, it is necessary
to reduce the background from scattered electrons reaching the detector. This is
accomplished in two ways:
the electrons but not the ions; and second, by pulsing the electron beam. The
ionizing beam is pulsed at 10 kc by applying a 40 volt pulse to the first lens of
the electron beam. The beam ''on'' time is 10 microseconds. After a delay of 5
microseconds, the second pulser sends a one volt pulse to turn on the transistor-
ized "AND" gate for 20 - 30 microseconds.
first, by establishing a weak magnetic field to deflect
Only those pulses which arrive at the
gate during this interval are sent on to the ratemeter. At the voltages used, most
of the electrons arrive at the detector vithin one or two microseconds; the flight
time of the negative ions, however, is of the order of 10 - 20 microseconds. Thus,
the signal to the ratemeter is primarily due to ions, with a drastic reduction in
the electron background. The output of the ratemeter is plotted versus the DC
voltage on the quadrupole with an XY recorder. A signal of five ions per second
can readily be observed. A typical plot, shown in Fig. 2, is the lover portion of
the mass spectrum of the negative ions of pentaborane.
All the boron hydride samples were purified by distillation in a vacuum
system. The diborane was trapped out at -154OC. The tetraborane was purified in
three steps: pumping off hydrogen and diborane while trapped at -130, passing
through a trap at -95 to condense the pentaborane and the higher boron hydrides,
and finally condensing at -196. The pentaborane was similarly purified: pumping
off diborane and tetraborane at -98, passing through a trap at -79, and final trapping
at -196.
Results and Discussion
The results are shown in the bar graph in Fig. 3. The narrow bars represent
peaks which have been multiplied by a factor of ten. All measurements are for 70
volt ionizing electrons. The monoisotopic mass spectrum was computed assuming the
natural BlO/B11 distribution of 20/80, and also assuming that there was no isotope
effect in loss of hydrogen from B10 versus B11. In nearly all cases, the Ykak
heights from which the monoisotopic spectra were calculated were an average of at
least three measurements, with separate measurements in agreement within 10%. In
calculating the monoisotopic spectrum, the system is "over-determined," :.e., there
are more equations than there are unknowns. This provides a means of checking the
internal consistency of the results. For every case except tvo, the self-consistency
deviation was less than 2%, and in all cases less than 10%. On the basis Of
these consi_derations, relative intensities within a group of peaks, e.g., BgHg ,
B5He , B5H7 , etc., should be accurate to 2 10%.
In comparing peaks between
different groups, however, it is more difficult to assign limits of error, for the
spectrometer may favor the observation of low masses over high masses in certain
cases. One can observe this "discrimination" in the positive ion spectra, and it
appears to be primarily a function of the resolution of the quadrupole. It is
possible, however, that the continuous electron multiplier and the ion source also
mass discriminate, perhaps differently for positive and negative ions.
Since the
signal from the multiplier is analyzed as pulses rather than current, mass discrim-
ination by the detector should be minimized. The exit holes in our ion source are
rather large, and it would be expected that discrimination here would also be
negligible. Thus, if it is assumed that the discrimination effects are the same
r
i,
f
1
L
r
,
4

A b 1 1
I H I
I I
FILAMEN T
CONTINUOUS
I
COWPR
SOURCE ION ' I QUAORUPOLE I
ELECTRON
SUPPLY I
AfULTIPLIER
I I
I
+
-
b
ELECTRON f QU*oauPoLE x-v POWER SUPPLY
ION OCTICS POWER SUPPLY RECORDER I PULSE
COWER SUCCLV f MASS SCAN AMPLIFIER
T
1 1 >
A -
A + t + 1
PULSCR I OSCl LLOSCOPE RATEME TER -
TRIGGER 4
OELAV
--------_
i CULStR 2
--_
i
GATE
4 4
FIGURE I
-

z"
0-
a?
Q
zc
c
rr) 'I

79
for positive and negative ions, approximate corrections can be made. At worst, the
relative intensities throughout the whole, spectra are probably correct within 225%.
Although reasonable precautions were taken to assure the purity of the sample
in the diborane spectrum, peaks corresponding to B3H7 and B4Hg were observed.
These could have arisen either from ion-molecule reactions, from pyrolysis of the
diborane as the sample flowed into the source, or from impurities i7 the diborane.
The latter is improbable, for in a positive ion spectrum of the same sample in the
same spectrometer, the tetraborane peaks could be observed only with difficulty.
It was estimated on the basis of the positive ion spectrum that the tetraborane
impurity could be no greater than one part in five-hundred. In the negative ion
spectrum the ?ea& corresponding to B3H7- and B2H3 were of almost equal intensity.
It should be noted, however, that in comparing the diborane and tetraborane samples,
it .was much easier to observe the negative ion peaks from tetraborane, which were
larger by sbmbnut two orders of magnitude.
There are several mechanisms by which negative ions can be formed. The tradi-
tional processes are as follows:
1. Ion pair. formation:
2. Resonance attachment: XY + e -+ XY
3. Resonance attachment with dissociation: XY + e -- X + Y-
XY + e -+ X+ +-Y- + e
In addition,a process somewhat like the third mentioned above can be imagined, in
which the neutral species is left in an excited state which in turn decays to an
ion pair. Also, a negative ion may fragment, giving rise to a neutral species and
a new negative ion.
4. Excited state decay: XYZ + e +. XY* + Z-
5. Fragmentation: XY- -+ X + Y-
XY* -+ x+ + Y-
+ Using the above scheme, and assuming that the fragments most easily lost are
BHg , BH3 , and hydrogens, most of the more intense peaks in the negative ion boron
hydride spectra can be accounted for rather well in the following manner.
Possible Modes of Fragmentation of Diborane
The underlined peaks are the four most intense in the diborane mass spectrum.
The other peaks come from loss of hydrogens, either singly or in pairs, from the
above species.
Possible Modes of Fragmentation of Tetraborane
This scheme accounts for all of-the peaks with an observed intensity of
greater than 1%, except for the B2H3 peak, whose intensity is 1.1%, and which could
be due to a diborane impurity.

80
Possible Modes of Fragmentation of Pentaborane
etc.
+
In support of the above mechanism, the data in Table 1 can be cited
BH2+
BH3+
Table 1
Abundances of BH2+ and BH3+ in Some Boron Hydrides
Diborane7 Tetraborano8 Pentaboraneg
19.1
.59
(These ion intensities were Cbtained with 70 volt electrons, and are given in
percent relative to the most intense peak in the mass spectrum for the cmpound in
quest ion. )
+
Thua, it can be argued that the loss of BH2 is much more likely than the loss of
BH3 . The loss of BH3, however, at least from neutral species, is fairly well
established in a number of reactions." It should not be too surprising if BH3 can
also be readily fragmented from negative ions.
- It is tempting to question the necessity of invoking the excited state decay
process (4) described above. For example, can the formation of the B3H7- not be
simply written as a resonant attachment process with the dissociation of BH3?
Although this is certainly possible, one or two points argue against this mechanism.
First, at the electron energies used, a resonant process of this type would not be
expected to be of primary importance. Secondly, one might inquire why the pair
ionization process,
~
+
4 + e ~ + B 1 ~ H ~ ~ + - B H ~
is not observed, especially in view of the fact that the analogous process seems to
be the predominant one in diborane. Indeed, the pair ionization process is the one
which is expected; it is disturbing that it is not seen in view of the stability of
B3Hg in solution and in the solid phase. The structure of B3Hg- is known frran
x-ray Work, and there is ample theoretical and experimental justification for its
existence.'l On the other hand, we have found no evidence in the literature for
B3H;. The question arises as to whether an error can be made in the mass assignments.
The marker compounds used were 02 and H20, giving rise to the 02 , OH-, and
0- peaks, which were always present in the background. Considerable care was taken
in making the mass assignments, so that unless some unknown instrumental or chemical
processes were occurring of which we are unaware, the given assignments are
correct. Therefore, granting that the reported observations are valid, and accepting
the notion that pair ionization should be $he dominant process at higher electron
enprgies, and further noting that the BH2 ion is much more predominant than
the BH3 in the positive ion spectrum, it seems necessary to invoke some mechanism
analogous to (4). Perhaps further weight can be given to this argument by noting
that the parent peak is not observed in tetraborane, suggesting that a loss of
hydrogen is a strongly favored process in the electron impact situation.
The present work suggests several further experiments which should be carried
7.0
.6
!
12.2
--

51
out to answer some of the questions which have been raised here. An investigation
Of metastable negative ions on a magnetic sector instrument should prove useful in
confirming or negating the fragmentation scheme postulated above. A careful study
of the diborane spectrum, either in a tandem instrument, or with an ion source in
which one could be sure that pyrolysis is not taking place would be necessary to
remove the ambiguity in the obseraation of tetraborane-in the diborane sample;
whether the peaks above diborane are due to impurities, or whether they do in fact
result from ion-molecule reactions. Perhaps most useful initially would be an
investigation of the boron hydride spectra at low electron energies, where resonance
attachment becomes the dominant process. Ideally, one should study the spectra as
a function of energy; the clastogram technique" might prove valuable in sorting
out the various processes occurring. We made some attempt to work at low electron
energies, but largely without success.
Since we are unable to measure the ionizing
current, i.e., keep it constant as a function of energy, we would have little con-
fidence in an appearance potential measurement, particularly since the background
from electrons again becomes a problem at low electron energies, when they can be
pushed out of the ionization chamber by the repeller.
Reese, Dibeler, and Mohler have reported on the spectrum of pentaborane, in
which they observed the resonant attachment process in the parent ion.13 Although
their relative intensities are different from those which we observed, this is
probably not surprising, in view of the different energies of the ionizing electrons
in the two cases. Curiously, however, Dibeler does not report any of the lower
fragments for pentaborane.
In conclusion, it is evident that the lower boron hydrides do form negative
ions. In certain circumstances, the abundance of these ions mag be great enough so
that they would have to be taken into consideration in mass spectral, radiation, or
gaseous discharge phenomena. It is also clear, however, that more than a mere qual-
itative study of these negative ions will be necessary for the complete understand-
ing of their roles in these situations.
Acknowledment- This work was done under the auspices of the United States Atomic
Energy Commission.
Figure Captions
Fig. 1. Block diagram of the mass spectrometer.
Fig. 2. Lower portion of the mass spectrum of the negative ions from pentaborane.
Fig. 3. The monoisotopic relative abundances for the negative ions frm diborane,
tetraborane, and pentaborane.
References
1. F. Fiquet-Fayard, Actions Chimiques et Biologiques des Radiations 11, p. 44
2.
(1965 1.
J. F. Paulson, Ion Molecule Reactions in the Gas Phase, p. 28, American Chemical
3.
4.
Society Publications (1966).
G. Jacobs and A. Henglein, Advances in Mass Spectroscopy 2, p. 28, The Institute
of Petroleum (1966).
M. F. Calcote and D. E. Jensen, Ion Molecule Reactions in the Gas Phase, p. 291,
American Chemical Society Publications (1966).
5.
6.
V.
V.
V. Subbanna, L. H. Hall, and W. S. Koski, J. Am.
Von Zahn, Rev. Sci. Inst. 34, 1 (1963).
Chem. SOC. 86, 1304 (1964).
7. MBS No. 333 Mass Spectra Data, American Petroleum Institute Project 44 (1949).
8. T. P. Fehlner and W. S. Koski, J. Am. Chem. SOC. &, 1905 (1963).
9. NBS No. 33b Mass Spectra Data, American Petroleum Institute Project 44 (1949).
10. T. P. Fehlner and W. S. Koski, J. Am. Chem. SOC., 86, 2733 (1964).
11. W. N. Lipscomb, Boron Hydrides, W. A. Benjamin, Inc. (1963).

82
12. R. W. Kiser, Introduction to Mass Spectrometry and Its Applications, p. 191,
Prentice-Hall, Inc. (1965). f
13. R. M. Reese, V. E. Dibeler, and F. L. Mohler, J. Research NBS 57, 367
I
I .
(
/ I
1956). ''
\I
J

93
INTERACTIONS OF EXCITED SPECIES AND ATOMS IN DISCHARGES**
*
Robert A. Young and Gilbert A. St. John
Stanford Research Institute
Menlo Park, California
ABSTRACT
The influence of atomic nitrogen on the concentration of N2(A3C:) excited in
nitrogen by a weak discharge has been studied. It was found that (1) a high-order
Process involving atomic nitrogen destroys N?(A3Ct), (2) a process utilizing
species, other than atomic nitrogen, created in the discharge removes Nz(A%$), and
(3) a process involving atomic nitrogen and another excited species produces
N~(A~c;).
It was also found that NO rapidly quenches N2(A3Ct)
with a rate coefficient
of 3 X 10-l' cm3/sec. A quarter of the N2(A3E:) destroyed excites NO to the
A~E+ VI = o level.
INTRODUCTION
Nitrogen subjected to an electrical discharge at low pressure has been a subject
of study for over 70 years. In general, phenomena occurring in the discharge itself
have been studied in static systems, while phenomena caused by long-lived species
generated by a discharge have been studied in fast-flow apparatus.
In all these phenomena there are apparently two classes of reactants involved,
atomic species and excited molecules. Recombination of atoms accounts for most of
the long-lived afterglow phenomena, while the latter are necessary for much that
occurs during the discharge and shortly (2 10 m sec) after its termination.
In both the discharge and the long-lived afterglow,' molecular nitrogen in the
A3E$ state is suspected of playing an important role. This is because it is
metastable, it is the lowest electronic state of N2, it is supposedly easily
excited in a discharge, and it is a necessary result of atom recombination. However,
emission from this state of N2, the Vegard-Kaplan bands, is exceedingly difficult
to observe in these phenomena. In fact the Vegard-Kaplan bands have never been
observed in the long-lived afterglow, and only under very special circumstances in
the discharge or in the short afterglow.
.) Recently' we attempted to observe the Vegard-Kaplan bands in the long-lived
(Lewis-Rayleigh) nitrogen afterglow. This afterglow is almost entirely due to the
first positive bands of N2 which terminate on the A3Ct state. The flux of this
radiation represents a minimum rate of production of N2(~3C+). From our failure
to observe the Vegard-Kaplan bands, from the radiative lifethe of the A3E: state
(5 10 sec) and from the sensitivity of detection, an upper limit to the actual
lifetime of N2(A3C$) in the afterglow of 5 x 10-4 sec was derived.
This result implied that collision processes were removing N2(A3Ct). However,
it is known that N2 does not quench N2(A3Et). Aside from N2, atomic nitrogen
is the largest constituent present, and we conjectured that it was responsible for the
short lifetime of N2(A3Ct).
**
This work was done under Contract DA-31-124-ARO-D-434.
* Visiting Fellow, Joint Institute for Laboratory Astrophysics, 1966-67.

84
This paper describes experiments aimed at investigating the hypothesis that
atomic nitrogen as well as several other compounds (NO, CO, NH3, H2), effectively
destroy N2(A3Et).
EXPERIMENTAL - ATOMIC NITROGEN INTERACTION
Nitrogen was excited by a pulsed or continuous 150 kc, 10 kV oscillator (of the
type used in high-voltage supplies) in a 5-liter quartz bulb with qimple one-eighth
inch tungsten rod electrodes separated by the diameter of the bulb. A one-half-meter
Jarrel-Ash Seya-Namioka monochromator was used to scan the discharge spectrum or to
isolate bands for decay measurements. Signal levels were very low, and an
Enhancetron signal integration unit' was used to abstract the decaying signal from
the noise.
Prepurified nitrogen was passed through a liquid nitrogen trap, then through
heated titanium-zirconium and two additional liquid nitrogen traps, past a microwave
excitation section, an NO titration inlet, and finally into the 5-liter observation
bulb, from which it was exhausted by a mechanical vacuum pump at approximately 100
cm3/sec. A filtered photomultiplier downstream from the bulb observed the N2 first
positive bands resulting from atom recombination.3 This permitted the atom concentration
to be computed after calibration against an NO titration.
RESULTS INVOLVING ATOMIC NITROGEN
Figure 1 shows spectrometer scans of the discharge. In Fig. la the N2 second
positive and NO Y bands predominate below 3000 A. The 6 bands were the only
other system of NO detected, and these were very weak. If the microwave discharge
is adjusted to produce a small quantity of atomic nitrogen, the NO bands can be
suppressed almost entirely (Fig. lb). Although the Vegard-Kaplan bands are also
reduced, the second positive bands remain relatively unchanged. Addition of NO
cacses a large increase in the intensity of the NO y bands (Fig. IC).
Figure 2 shows the decay of the (0.5) Vegard-Kaplan band and the (0,3) NO
band with the addition of atoms, &, when the upstream microwave discharge was on.
The decay of the NO Y bands is identical to that of the Vegard-Kaplan bands.
Figure 3 demonstrates the buildup and decay of the (0,5) Vegard-Kaplan band
when the 150 kc discharge is turned on and off; the rise and fall of this band
involved the same exponential.
Figure 4 indicates the rate of decay of the Vegard-Kaplan bands as a function of
pressure without added nitrogen atoms.
Figure 5 shows the dependence of the Vegard-Kaplan decay rate on [NI2 derived
from the downstream photomultiplier.
Figure 6 illustrates the pressure dependence of the rate coefficient for
de-excitation of N2(A3Z$) by atomic nitrogen.
Figure 7 is a plot of the growth rate constant of the Vegard-Kaplan bands,
after the discharge is turned on, as a function of [NI2.
Figure 8 shows (Io/I)vk as a function of atomic nitrogen, where IoVk IS the
intensity without added atoms and where Ivk is the intensity vith atoms.
Figure 9 indicates the decreased number of nitrogen atoms leaving the quartz
bulb due to the discharge.
.

L
4
I*C z
0'2 =
a,
a
U x
I
m
4
m
0
a,
CI
1'0 =
m
0'1 =
C'C =
1'2 =
I'I =.
0'0 =
t'Z = -3
C'I =
2'0 =
S't z
C'I =
C'O =
C'O -
L'E z
9'0 =
6'1 5
x
0
8 M
r
U
d
e
71
d
w
d

Pig. 2
z
w
l-
4
W
a
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.2
0.1
-.
- PRESSURE 0.8m m Hg
-
-
-
-
VEG ARD-K APL A N
0toms/cm3 add(
I I I
0 10 20 30
TIME- rnsec
10-452522-40
_-
The intensity of the (0.5) Vegard-Kaplan band and the (0.31.~ band.as a
function of time for various amounts of added atomic nitrogen.
0
0

I
L
I,
;
>
>
E
Fig. 3
I50
200
150
100
50
. .
s;
0
msec 0 25 50 75 100 125
TA- 45 2 5 2 2- 4 2
Buildup and decay of the (0,5) Vegard-Kaplan band when the 150-kc discharge
is turned on and off. The curve represents integration of approxirtely
1000 cycles.
-1
-
0 " 0 '
1
I
0 0.5 I .o 1.5 2.0 2.3
1
i
,
? .
PRESSUSE-nmHg
..
0
TB-452522-41
P Fig. 4 The dependence of the rate of decay of the Vegard-Kaplan bands as a
function of pressure without added n.itrogen atoms.
I
A*>
I
i

1' &p
400 2.0 mm
300
100
0 4 0 " 5 10 15 20
ATOMS [NI2 x (atoms/cm3)
ADDED
T0- 452522-47 ~
Fig. 5 The rate of decay of the Vegard-Kaplan bands as a function of the square
of the nitrogen atom concentration. 4
-
20
15
IO
5
0 0 0.5 1.0 . 1.5 2.0
PRESSURE-~~H~
TA-452522-44
Fig. 6 The ressure dependence of the rate,coefficient for atom destruction of the
N3(A s Et) state.
2
-- -
i
<
/
i
1

I 300
r
1
- 200
i io
;' a
cn
I I.
0 5 IO 15 20
I . -
Fig. 7 The buildup rate of the Vegard-Kaplan bands during the excitation discharge
as a function of the square of the atomic nitrogen concentration.
4
\
I
I .
2.50
2.00
1.50
[N] (otoms/crn3)
0 0.5 I .o 1.5 2.0 2.5~10'~
- I I I I -
-
1.00
2 3 4 5 x1026
/ [NI2 (a toms /c m3I2
-
T8-452522-49
The. normalized inverse intensity of the Vegard-Kaplan bands during the
excitation discharge as a function of added atomic nitrogen. The discharge
was run continuously.

E
E

where
DISCUSSION OF XESULTS PERTAINING TO ATOMIC NITROGEN
The observations indi,cate that duriyg decay
1 1
- = -
T
0
+ K [NI2 [MI
with K 2 4 x cm9/sec. Neither or can be the radiative lifetime of
N2(A3X3 'which is 210 sec.
The differential equation describing the time dependence of
3+
d[N2(A Eu) 1
dt
= ' P - L. [N2(A 3 Eu)] +
I
3+
[N2(A Zu)] is
3+
where L contains all factors other than [N2(A XU)] such as collision partner
concentrations, rates of radiation or diffusion, s., that are involved in processes
removing N2(A3Et), and where P is the rate of production. Substituting Eq. (1)
in Eq. (3) gives
or
Then either
3 + -t/.'o 1
[N2(A Eu) lo e (L--) = P .
(4)
'a
1
L = -
Ta
P
1 3+
(L - [N2(A Xu) lo
a
(3)
and P = O (5)
-t/Ta ,
= e (6)
-t/Ta
In the latter eventuality L > l/ia, since neither P nor e can be negative.
If L is comparable to 1/~~, it must be essentially constant in time,' while if it
is large compared to l/ia,
Since all constitutents of the system probably decay when the power input is
stopped, would be characteristic of the decay of a reactant concentration
ia
involved in P to a higher power than in L, while other reactant concentrations
are either constant or common to both P and L. The previous statement would be
true if a product of reactant concentrations replaces a single reactant concentration.
Since 1 / changes ~ ~ by almost a factor of 10 over the range of [N] used,
either L >> 1/~, for small [N] and hence P/L [N2(A3E;)lo = e-t/Ta, or L

92
closely follows 1 / in ~ dependence ~ on [N]. *, L a l/Ta. Discussion of the
~
last possibility will be deferred. In the former case L must be very large indeed
and [N2(A3xt)] must be essentially in a quasi-steady state with its production
rate, & T~ refers to the decay of P. Since [N2(A3Et)] is not strongly
deactivated by pure N2, L would certainly not increase when the energy of the
system dissipates as would be necessary if P did not decay. Because the decay of
[N2(A3c:)] appears to remain exponential for all values of [N] and because Ta
is a smooth function of [N] then L > 1/T (max) if the data are; interpreted as
the decay of P with [N2(A3xi)] in a quasf-steady state. For this to be true
some species capable of deactivating N2(A31t) must be generated in the discharge
and remain constant for times long compared to those where [N2(A3C$)] is detectable.
The most abundant chemical species generated by the discharge is N; the most
abundant excited molecule is Nz(A3L:) -- except possibly for the 3A state of N2 I
which has not yet been directly detected. Only atomic nitrogen appears to be a
realistic choice for the species which decays slowly and would keep L >> l/Ta.
i.
I
. I
However, a difficulty arises in the interpretation with L >> l/Ta if we consider
the observed behavior of [N2(A3 :>I when the discharge is turned on. The rise of 1
[N2(A3xz)] to a steady state is describable by
If this is substituted into Eq. (3). we obtain
3 +
If -1/~~ = I,, then P = L [N2(A E,)] = Po, a constant in time. When t = 0,
P = Po = ([N2(A3~~)]o/Ta) if L # (19~~); and if P = Po + P', we obtain
If L >> 1/~~, P' is positive; and if L > 1 / ~ 1 ~
is the only one tenable during the excitation discharge. 4
Since Po follows the discharge transients, it must become zero when the dis-
charge is turned off, so that P changes rapidly compared with Ta from P + P' 1
to P'. But this is inconsistent with the interpretation of the decay of [82(A3$)1 ,
previously given on the assumption that L >> l / because ~ ~ this rapid fall in P
would be reflected in a rapid fall of [N2(A3C$)] which is not observed. Hence this
contradiction forces L = l/ta and P = Po, a constant, during the discharge and I
L = l/ra and P = 0 after the discharge.
During the excitation discharge a steady state Is reached, and
j
>
I
!
I
4

93
where the bar indicates values constant in time. Then indexing all quantities with
a subscript 1 when atoms are not added and with N when they are added, we obtain
where the last equality uses the results of Fig. 8. Hence 4 involves [N] to a
higher power than does sN. Thus T~ cannot refer to the decay of [Nl. This is
consistent with the usual slow decay of [N].
Since 4 is essentially proportional to [NI2, these results imply that FN
is proportional to [N]. If this additional production of N2(A3Z$) utilizes
nitrogen atoms on a one-for-one basis, [N] must decrease in passing through the
discharge, &,
where At is an average residence time, where L[NlOut represents the rate of atom
removal, and where mixing is rapid compared with reaction. This leads to
or
which is the general form observed.
From Fig. 9 we can obtain L (using At %50 sec), and from Fig. 8 with a slope '
of (KbPo[N2]/L), we can also obtain L if KT,P,[N~] can be found. A comparison
of L derived from Fig. 9 and Fi . 8 is then possible. Since P = N2(A3L;fi>]/~,,
absolute measurements of [N2(A3$)] derived from I(VK) = ( [N2(A3~)]/Tr) L where
is the radiative lifetime of the excited state give T~P.
T~
The absolute intensity of the 0,5 Vegard-Kaplan band in the discharge bulb can
, be roughly obtained by measuring its signal and calibrating the spectrometer response
using the NO y (0,3) band excited by the chemiluminescent association of N and 0
in conjunction with measurements of [N] and [O] and the known rate coefficient for
I these pro~esses.~ In general, the comparisons of L derived from Figs. 8 and 9 show
' that if nitrogen atoms are consumed in the excitation of N2(A3c), then several
I hundred times more atoms need to disappear than in fact are destroyed. This implies
I that nitrogen atoms are not consumed but catalyze the production of N2(A3~t) in
conjunction with other species generated in the discharge.
> A similar argument can be used to show that atomic nitrogen is not consumed in
1 destroying N2(A3c). For example, there are often as many N~(A3zt)~ $olecules as
\ N atoms; these would be seriously depleted during the decay of N2(A zU) and thus
would lead to a nonexponential behavior.
,
i
'1
3 +
The short lifetime of N2(A &,> in the discharge (Fig. 4) cannot be due to atoms
because their concentration, as inferred from downstream chemiluminescence measurement,
' is too small. Furthermore, the pressure dependence of T~ is incompatible with atom

94
deactivation. Hence, some species is produced in the discharge in addition to atomic
nitrogen which is capable of rapidly deactivating N2(A3X$). This species persists
for times long compared to T . '1
The results of these experiments indicate that 3everal unusual processes are
occurring: (1) a high-order process destroying N2(A 1:) which involves atomic
nitrogen; (2) a process utilizing species, created in a discharge other than atomic
nigrogen, which removes N2(A3X$); and (3) a process involving atomic nitrogen and
another species generated in the discharge which produces hi2 (A31i)'. These processes .
are in addition to electron excitation and radiative destruction of N2(A3C$). Several
eossible reaction schemes can be constructed which are consistent with our observations.
However, without positive identification of all the species created in the discharge
and the istermediaries necessarily involved in high-order processes, these reaction
schemes are speculative. It seems best to postpone discussions of these until more
experimental information is available.
EXPERIMENTAL REACTIONS INVOLVING NO
For experiments on the effect of NO on N2(A
3 Zh), three photomultipliers were
installed downstream from the RF excitation bulb to observe chemiluminescence. One
photomultiplier had an interference filter centered on the strong bands of N2 at
5800 A, a second (called the yellow detector) observed radiation at wave lengths
longer than 5200 A, and a third (called the ultraviolet detector) observed emission
between 3800 A and the pyrex cutoff. The first two photomultipliers are sensitive
to the N2 first positive bands excited by recombination of atomic nitrogen and to
the NO2 continuum excited by reactions of 0 with NO, while the ultraviolet
photomultiplier is mainly sensitive to the NO, 6 bands excited in the association of
atomic nitrogen and oxygen.
These photomultipliers were calibrated by generating
nitrogen atoms in the microwave discharge and titrating them with NO. It should be
remembered that all emission observed in the downstream bulb is of chemiluminescent
origin.
RESULTS
Figure 10 shows the decay rate when the exciting discharge is turned off from the
0,3 NO y band, and the 0,5 Vegard-Kaplan band as a function of atomic nitrogen added
by the upstream microwave discharge.
Figures 11 and 12 show the intensity of the 0,5 Vegard-Kaplan band, the 0,3 NO y
band and the 0,6 NO 6 band in the discharge bulb as a function of the NO concentration
that would have existed in the stream if no destruction of NO occurred. Also shown
are the responses of the yellow photomultipler and the ultraviolet photomultiplier,
both of which are located at the downstream bulb.
Spectrometer scans of the discharge without adding NO and on the plateau of
Figs. 11 and 12 are shown in Fig. 1.
Figure 13 is a plot of the area intensity ratio of 0,3 NO y/0,5 Vegard-Kaplan
bands as a function of NO added beyond the null and as a function of the relative
response of the yellow phototube to the NO2 continuum. Figure 14 is similar to Fig-
13 but involves the 1,lO Vegard-Kaplan band and the 3,O NO y band.
DISCUSSION
The simultaneous decay of N2(A
3
Xu)
+
and Iys the intensity of the NO y bands,
suggests that NO is excited in the process5
NO(X 2 ll) + N2(A32:) -f NO(A22+) + N2(X12) .
1
,
i

'I
i
'I
0 2. 4 6 8
2
[ N]* x' I 0-26 (otorns/crn3)
"
IO
TB-- 45 2 5 22- 37
Fig. 10 The dependence of the rate of decay of N2(A 3 xu) + and NO(A 24- I ) as a function
of the square of the atomic nitrogen concentration.
IO0
80
GO
40
20
0
0 IO 20 30 40 50 '60 70 80
[NO] x IO-" (molecules/cm3)
TO-452522-39
11 The dependence of emission intensities on NO concentration: a- the NO B
bands, A - the N2 first positive bands, 0 - the NO2 continuum -- all excited
in the downstream observation bulb; 0 - the 0,3 NO y band, A - the 0,6 NO 6
band, and 0- the 0,s Vegard-Kaplan band excited in the Tessla discharge
bulb *

[NO] x IO-" (molecules/cm3) .
18- 4 52 5 22-48
. . .
---
Fig. 12 The dependence of emission intensities on NO concentration: m - the NO B
bands, 0 - the NO2 continuum -- both excited in the downstream observation
bulb; 0 - the 0,3 NO y band, A - the 0.6 NO f3 band; and 0- the 0,5
Vegard-Kaplan band excited in the Tessla discharge bulb.
[NO]x IO-" (rnolecules/cm3)
20 22 24 26 28 30 32 34 36
I500
PRESSURE 1.8 mmHg
".
0 O-
A 0
" 5
H
2,
Fig. 13
0
SLOPE = I. I 10-9~m3
7 21 35 49 63 77 91 105 119
RELATIVE RESPONSE YELLOW PHOTOTUBE (NOZcontinuum)
-
78-452522-36
The relative photomultiplier response to the NO y (0,3) and N2(0,5) Vegard-
Kaplan bands as a function of both [NO] and the response of the yellow
photomultiplier.

Fig. 14
IO
8
6
4F
[NO] x IO-" (mole c u I es /c m3)
14 16 18 20 22 24 26
I
SLOPE = I x 10-11cni3
0 IO 30 50 70 90 110 130
RE L AT I V E RES PO N S E YE L LOW P H OTO T U B E ( N O2 con t i n u u m )
T8-452522-51
The relative photomultiplier response to the NO y (3,O) and N2(1,10)
Vegard-Kaplan bands as a function of both [NO] and the response of the
yellow photomultiplier.

O-‘
The initial decrease of the signals from the yellow downstream photomultiplier
is due to NO titration of nitrogen atoms produced in the exciting discharge, &,
The null point represents the complete conversion of nitrogen atoms t;o oxygen atoms.
The concentration of these atoms derived from the amount of NO added to reach this
null is consistent with the intensity of the first positive bands before NO addition.
The linear rise of the yellow sensitive photomultiplier after null is due to
and indicates that 0 is essentially constant.
NO + 0 -+ NO2 + hv (18)
There is a slow increase of the NO y and f3 bands in the excitation bulb because
of the slow increase in the actual steady-state concentration of NO which is main-
tained by
and reaction (16), a,
N + 0 + M -+ NO + M
NO+O+M+N02+M
NO + 0 -+ NO + O2
When [N] goes to zero, this steady state cannot be maintained, and NO in the stream
increases as expected for no reaction of NO. In the reaction time available reaction
(20) does not significantly alter either [O] or [NO].
As [NO] increases sharply after the removal of N, the NO y and f3 bands increase
sharply in the discharge and the N2 Vegard-Kaplan bands decrease rapidly. Since
I Y = [NO(A2E+)]/r,
for excitation in the discharge bulb, and since
(22)

for chemiluminescent excitation in the downstream bulb, reaction (16) implies
49
3+
I = kl6[NO1 [N2(A Zu) 1 = k16[NOl Ivk~vk =
Y
2 +
if NO(A C ) is only deactivated by emission.
Since 0 is constant,
This is true, as Figs. 13 and 14 show.
but relative.
k161N02 'vkTvk
In these graphs intensities are not absolute,
To obtain k from Eq. (26), we must experimentally relate the emission
intensities more agrectly to the concentration of the emitting state. If Iy(0,3)
represents the absolute intensity of the 0,3 y band of NO and similarly for the
Vegard-Kaplan bands, then Eq. (11) becomes
where fy(0,3) is the fraction of all emission from the v' = 0 level which terminates
on the v" = 3 level and similarly for f (0,5) and where T is the radiative lifetime
of the A32+ state6 and 6 representsvhe fraction of allv% (A313 which are in
v' = 0. Only vy = 1 and 0 are detected by their emission, an2 6 is found to be
one half and constant. Then
The parameters in parentheses depend only on the internal characteristics of the
molecules and have previously been measured6 as
fy(0,3) = 0.14. Hence
T vk = 20 sec, fvk(0,5) = 0.17, and
0.03 I (0,3)
k16 = [NO]Iv:(0.5) '
To obtain k , only the relative intensity of Iy(0,3) and IVk(O,5) are
required. Since kkese bands occur very near each other in the spectra of the discharge,
their relative intensity is equal to their relative signal level. Hence k16 is 0.03
times the slope of the line in Fig. 13, &,
k16 = 0.03 (1.1 x lo-') = 3 x cm3/sec . (31)
7
This rate is much larger than that reported by Dugan.

100
The existence of the 6 bands and emission from higher vibrational levels of the
NO(A2Z+) state indicate that this rate is a lower bound to the rate of energy transfer
from N2(A3Z$) to NO(X2JI). There are several states of NO
energy exchange involving N2(A3C+) -- the a4n, b4n, b4Z-,
accessible in an
as well as the A2Z+
and B2n which are revealed by deir emission.
The total rate of de-excitation of
3+
N2(A Zu) can be estimated, since we know,
in the absence of NO, its deactivation rate is 1/~ = 50/sec, and the intensit
of the Vegard-Kaplan bands is reduced by half when [NO] % 7 x 10l1 &olecules/ cm 3 ,
i.e., 50 = ki6(7 x 1011); in other words
ki6 = 7 x cm3/sec
3+
or essentiallya quarter of all the N2(A 1 )
U
v' = 0 level of the NO(A2Z+).
deactivated leads to excitation of the
2
It is apparent from the spectra that excitation of the A Z state of NO by
Nz(A3Z:) decreases in efficiency as v' increases in the NO molecule, becoming
very small for v' = 3 which is the highest level observed. This level can only be
excited by Nz(A3Zt v = l), while the v' = 2, 1, and 0 levels of the NO(A2C+)
state can be excited by
the rate coefficient of
N2(A3C$ v = 0,l). Using the data in Fig. 14, we can obtain
as
1.
2.
3.
4.
5.
6.
7.
1 2 +
N2(A31: v' = 1) + NO(X2n) -+ N2(X Z) + NO(A Z v = 3)
k (1,3) = 10-l~ cm3/sec. .
16
REFERENCES
R. A. Young, Canad. J. Chem. 46, 1171 (1966). This paper contains additional
references. K. Wray independently arrived at similar conclusions in J. Chem.
Phys. 44, 623 (1966).
Enhancetron Nuclear Data Inc.
R. A. Young and G. Black, J. Chem. Phys. 44, 3741 (1966). This paper has references
leading back to the description of most of the phenomenon associated with dis-
charged nitrogen.
R. A. Young and R. L. Sharpless, Discussions Faraday SOC. 33, 228 (1962).
A. B. Collear and I. W. M. Smith, Trans. Faraday SO~. 6l, 2383 (1965). .,-
W. R. Brennan, Studies in active nitrogen, Ph. D. Thesis, Dept. of Chemistry,
Harvard University, Sept. 1964, and references quoted here; R. W. Nicholls,
J. Res. Natl. Bureau Std. (U.S.) m, 451 (1961); R. A. Young and R. L. Sharpless,
I
J. Chem. Phys. 39, 1071 (1963); R. A. Young and R. L. Sharpless, J. Chem. Phys.
- 39, 1971 (1963); R. A. Young and G. Black, J. Chem. Phys. (in press). These
references contain an extensive review of the pertinent chemiluminescent reactions. !
C. H. Dugan, An experimental study of metastable atoms and molecules, Thesis,
Division of Engineering and Applied Physics, Harvard University, June 1963;
C. H. Dugan and N. P. Carleton (in press).
(33)
I
1

101
Chemiluminescent Reactions of Excited Helium with' Nitrogen and Oxygen
Mark Cher .
North American Aviation Science Center, Thousand Oaks, California 91360
Abstract
Active sp,ecies created in a fast flov of helium by a micrdwave discharge
react outside the $ischar e with nitrogen or oxygen'causing the emission of
9
visible bands of N. and 0 . .
The emission intensity of thesebands decays '
exponentially with'distanze,' and the decay coefficient depends on both the
flow rate of helium and the flow rate of nitrogen or oxygen. A t constant
flow rate of helium the decay coefficient increases with increasing flow rate
of added gas, but the dependence is less than linear. A theoretical analysis
of the flow system predicts the exponential decay and a linear dependence
between the decay coefficient and the flow rate of added gas. Comparison of
the experimental and theoretical results permits the calculation of approximate
values of the rate constants for the reactions populating the emitting
states and the diffusion coefficients of the excited helium species. A
plausible explanation and a means for removing the discrepancy between the
experimental and theoret.ica1 results are suggested.
In t r oduc t i on
Long lived reactive species, produced when helium gas is subjected to an
electrical discharge, react with many other gases and generate characteristic
emissions of visible light. The nature of some of these che i uminescent
reactions'has been discussed recent.ly in a number of papers. e-3 The reaction
with nitrogen gas produces+an intens2 !right blue flame consisting of the
2 +
first negative system. of N (B-C X C ). With oxygen a bright yellow-green
flamt is obeerved due to tze ex!itatiofi of both the+fisst neggtiv,e system of
Oi(b 2- -. a n ) and the second negativ,e system of 0 (A n -. X n 1. In this
paper Be repoyt the results of our measurements of the sFecial gariation of
the emission intensity in a fast flow system for various flow conditions,and
show how these measurements can be used to evaluate the rate constants for
the reactions populating the emitting states.
Apparatus. The reaction cell and associated equipment are shown schematically
in Fig. 1. The reaction cell consisted of a pair of concentric
pyrex tubes 13 and 25 mm 0.d. Helium flowed through the inner tube. The
titrating gas was introduced into the helium stream via the outer tube
through eight 1 mm diameter holes located sytmetrically in the wall of the
inner tube. High velocity flows of about 10 cm/sec were maintained by a
mechanical vacuum pump rated at 425 liters/sec. The flow rates of the gases
were measured using a pair of calib ated critical velocity orifice flowmeters
2
described by Andersen and Friedman. The pressure in the reaction cell was
taken as the average value measured by two oil manometers located approximately
20 cm upstream and downstream from the flame zone. The temperature
was determined in separate experiments by means of a fine wire thermocouple
sealed from the outside with black wax. The observed temperature was determined
primarily by the pressure of helium in the discharge, and was nearly
independent of the nature and flow rate of the added gas.
The discharge $n the helium stream was excited by a 2450 Mc/sec cavity
of the Evcnson type, powered by a 125 W Raytheon diathermy unit. The cavity
was located 85 mm upstream of the injection holes. A metal shield excluded
the active discharge from extending into the flame zone.
Spectra and intensity measurements were made with a 500 mm Jarrell-Ash
Ebert spectrophotometer using an.RCA 1P21 photomultipl.ier tube. A pair.of
large condensing lenses served to focus the image of a cross-section of the
flame on the entrance slit. The spectrophotometer was mounted on a table
which could be moved at constant speed parallel to the axis of the tube. In
this way the intensity of-the flame was recorded as a function of distance
along the tube. i

1.8
0.21
I
Titrotion with Np 01 3915A
9.30
MOlor diiven loble
I I I I 1 I
0 0.5 1.0 1.5 2.0 2.5 3.0
F,,~ I io6 (rnoleslsec)
5
c
1

103
All gases were taken directly from the commercially available high pressure
cylinders. The stated concentration of impurities in the helium tanks
was 50 ppm, with N2 being the major contaminant.
Interpretation of Data
We assume that an excitTd heliyp) species X* reacts with N or O2 to give
the electronically excited N or O2 , and the Latter immediately decays by
emitting a quantum of light,2as indicated by reactions (1) and (2)
8 +*
X + N2 - N2 + x
Na* - N+ + hv (2)
2
At this point the id ntity of X remains unspecified. X could be for
example metastable 2 9 S helium atom, in whi$h case X represeqta a ground state
He atom plus an electron.
Alternatively X could be the He2 molecule, and X
would then represent two ground state helium atoms. For simplicity we assume
that under any one set of conditions only one excited helium species is
dominant, although we can gxpect+geveral processes occurring simultaneously.
The natural lifetime of N+ or 0 is short compared with the time scale of
the experiment, and thus ?he emigsion intensity is proportional to the rate
of reaction (I). If the concentration of nitrogen is uniform and constant
along the tube, which should be the case after some distance of travel to
effect complete mixing, the decgy in intensity with distance should parallel
the decay in concentration,of X . ne therefore need to derive an expression
f,or the concentration of X as a function of position.
Ue consider a semi-infinite cylinder of radius r . Let n be the concentration
of X and N the concentration of the addadogas, say N2. We assume
tbat the two major mechanisms responsible for the decay of n are diffusion of
X to the walls where n=O and the chemical reaction represented by Eq. (1).
Under these conditions the rate of change of n is given by
u = Dv2, - kNn (3)
ax
where u is the stream flow velocity, x is the axial coordinate, D is the dif-
fusion coefficient of X , and k is the specific rate constant for reaction (1).
Since transport along the axial direction by convcctio~~ is much greater than
by diffusion, we need only consider radial diffusion, and for this case the
solution of the axial part of Eq. (3) is
A, the characteristic diffusion length given by A = r /2.405, is obtainedfrom
the solution of the ra la1 part of Eq. (3) using the gssumed boundary condi-
tion n=O at the walls. The radial dependence of n(x) is taken care by
the experimental arrangement, since the photomultiplier tube views an entire
cross-section of the flame, and its response is proportional to the average
intensity. The flow velocity u and the concentration N are calculated by
Eqm. (5) and (6)
u = RTZF/pWro 2
(5)
N = (F,/IF)(p/RT) (6)
where p is the pressure, T is the temperature, FN is the flow rate of the
added gas in molee/sec, and ZF is the total flow rate, which in these experi-
ments is essentially equal to-the flow rate of helium. If the emitted inten-
aity I(x) is proportional to n(x), it follows that a plot of An I c. x should
be a straight line with slope
2
(Dopa) (2.405)2n pr
--zs din1
dx = 760 XFRT + kW &&> FN (7)
~ ~.
10 Eq (7) we make use of the fact that D is inversely proportional to pressure
and (D p ) is the diffusion coefficient at pressure 1 mm Hg. Using Eq (8) it
0 0
-

should be possible to evaluate k and D p from the variation of S with flow
0 0
rate of added gas.
Results
veasurements of intensity VS. distance were recorded for nitrogen at
3915 A and for oxygen at 5587 G n d 4116 8. These wave length? are the heads
of intense vibrational bands of $he first negative system of N and the first
2 .
and second negative systems of 02. Typical flames were 2-10 cm In length,and
over this distance the intensity varied by nearly a factor of 100. Plots of
the logarithm of intensity x. distance demonstrated a linear bqhavior over a
50 fold change in intensity, as predicted from Eq (7). Negative deviations
from linearity occurred in the region within 0.5-1.0 cm from the injection
ports, this no doubt resulting from incomplete mixing of t he reactive gases.
The slopes of the linear portions were plotted against the flow rate of the
added gas, and the results are shown in Figs. 2-4. It is evident that theae
curves are not linear over the range of flow rates covered, as expected from
Eq. 8, but rather show a pronounced downward curvature. We have provisionally
taken the extrapolated intercepts and the initial slopes and used them to-
gether with Eq. (7) to compute the reaction rgte constant as well as the
diffusion coefficient of the excited species X . The results are shown in
Table I. The magnitude of the rate constants fns the nitrpgen-fnd oxygen
reactions are comparable and in the range of 10 cc mole sec . These rate
constants are very fast indeed being of the order of the colli6ion frequency.
The diffusion coefficients ?re of the right order of magnitude for any one of
the excited helium species, but the observed temperature dependence is
clearly too steep.
Discussion
The procedure of using initial slopes to evaluate the rate constants in
Table I is admittedly questionable in view of the fact that the curves in
Figs. 2-4 do not fit well the model embodied in Eq. (8). Although the intensity
of the flames does decay exponentially with distance, and the decay
coefficient increases with increasing flow rate of added gas, the increase is
not 1-inear. Deviations from linearity appear to be more pronounced at higher
pressures (i.e. higher flow rates of helium), presumably because of slower
mixing of the reactive streams. This suggests that incomplete nixing and
non-uniform concentration of added gas may be responsible for the breakdown
of the model, even though the method of injecting the second gas into the
helium stream with an initial radial component of velocity should encourage
fast mixing. At the lower flow rates of added gas the intensity of the flame
decays less rapidly, and consequently intensity measurements are made farther
away from the inlet of the second gas. Under these conditions we expect
mixing to be mora nearly complete and this is in fact the rationale for using
the initial part of the curves for evaluating the rate constants. Ye can
check whether incomplete mixing is indeed the source of our difficulties by
changing the size of the inlet holes and seeing whether the intensity profiles
are affected. We are currently carrying out these experiaentm.
we have not consideref so far the identity of the excited hflium specie8
x Collins and Robertson have shown that the upper state of N giving ri e
to the blye omission is populated by reaction of N with both ae$astable 2’s
2
He and He . +Similarly they have shown that the upper states of both band
systems of Oz are populated by reaction of 0 with 2 3S He,+and in addition
2
the upper state of the 5587R system is also populated by HeZ. Difference8 in
the O2 titration curves for the two band systems, paTticularly with regard to
tho diffusion coefficients at high pressures when He2 is dominant, may result
from the fact that different species give rise to the emission. More accurate
measurements of the diffusion coefficient are necessary to resolve this point.
There are very few measurements of the rate constants for th se reactions
8
with which we may compaft our resu ts. Pehsenfeld and co-worRers report a
rate constant of 4 x 10 cc -1
mole sec for reaction (8)

10;
He: + N2 -. 2He + h': (8)
whi!? Sholette and Muschlitz' iive values of 5-x 1013 and 11 x 1013 cc mole--'
sec for the reactions (1) and (10)
He(23S) + N2 - He + N+ + e-
2
He(Z3S) + O2 -.. He + 06 + e-
(9)
( 10)
Our preliminary results are in quite reasonable agreement. 1
References
1. C. 8. Collins and W. W. Robertson, J. Chem. Phys. 2, 701 (1964); 40, 202
(1964); lo, 2208 (1964).
2. A. L. Schmcltekopf and H. P. Broida, J. Chem. Phys. 2, 1261 (1963).
3.
4.
5.
E. E. Ferguson, F. C. Fehsenfeld, P. D. Goldan, A. L. Schmeltekopf, and
H. I. Schiff, Planetary Space Sci. 2, 823 (1965).
J. W. Andersen and R. Friedman, Rev. Sci. Just. 2, 61 (1949).
F. C. Fehsenfeld, K. M. Evenson, and H. P. Broida, Rev. Sci. Just. 96,
294 (19651.
6. E. W. HcDaniel, Collision Processes in Ionized Gases (John Wiley & Sons,
Inc., New York, 1964), p. 503.
7. Ibid, p. 516.
8. F. C. Fehsenfeld, A. L. Schmeltekopf, P. D. Goldan, H. I. Schiff, and
E. E. Ferguson, J. Chem. Phys. 5, 4087 (1966).
9. W. P. Sholette and E. E. Huschlitz, J. Chem. Phys. 2, 3368 (1965).
a.
Run
0
Table I
Titration with N2 at 3915 A
4
x 10
FHc
moles/sec
P
mm Hg
T
OK
k x
cc mole-lsec-l
2
(cm scc )(mm Hg)
1 6.93 2.70 363 2.1 374
2 9-30 3.46 376 2.0 480
3 16.1 ' 5.41 413 2.8 696
4 30.6 9.64 480 4.2 2140
5 39.6 12.1 508 4.4 3310
b. Titration with O2 at 5587 (First negative system of 0') 2
6 7.94 2-99 370 6.7 40 3
7 16.1 5-36 409 8.4 543
8 30.8 9.58 475 9.0 1660
9 45.6 13.8 515 8.3 4030
c. Titration with O2 at 4118 (Second negative system of O*)
2
10
11
12
13
8.00
16.1
30.5
45.5
3.11
5.40
9.58
14.0
3 66
418
488
5 26
5.7
5.3
6.4
5.1
33 1
643
1020
2140
f

THE MATHEMATICS OF STEADY-STATE DIFFUSION AND FLOW TUBE SYSTEMS
11. Measurement of Discharge-Zone Rate Parameters
Peter R. Rony
Central Research Department
Monsanto Company
St. Louis, Missouri '
I. INTRODUCTION
The techniques used to measure the rate of a chemical reaction or to
determine the physical characteristics of a chemical compound are dictated by
the physical and chemical properties of the compound, particularly by its lifetime
and reactivity in relation to its chemical environment. This environment
can be chosen to maximize the interactions between the compound and other
materials (a reaction environment) or else to minimize such interactions (an
isolation environment). Progress in chemistry has been dependent upon the
skill chemists have shown in selecting the appropriate reaction or isolation
environments.
The problem of developing suitable isolation environments has been a
difficult one in the study of flames, shock waves, explosions, and electrical
discharges, where the chemical intermediates -- ions, electrons, atoms, free
radicals, excited atoms, excited molecules, excited ions, and metastable mole-
cules -- are highly reactive and have extremely short lifetimes. Notable
advances along these lines have been the matrix isolation technique developed
by Pimentel and co-workers, contacting the intermediates with a very cold
surface, and the use of materials that deactivate surfaces against the recom-
bination of atoms and free None of these techniques have yet
been successful in significantly prolonging the lifetime of an ion.
This problem can be circumvented by the expedient of placing a source of
the chemical intermediates in close proximity to a sink and optimizing the rates
of loss and transport of the highly reactive species between these two regions.
The most popular source-sink system for the study of gas-phase kinetics of neutral
molecules is the diffusion or flow tube, which usually employs an electrodeless
discharge as the source.s,io
Such systems have also been used with considerable
success in the fieId of plasma chemistry for the experimental measurement of
associative detachment and ion-neutral reaction rates.11,12
The simplicity of these systems is deceptive -- they are usually quite
complex. To illustrate this point, Table I gives a non-exhaustive list of
rate processes occurring within the system shown in Fig. l.24 It is clear that
a comprehensive mathematical description of such systems is desirable not only
to aid in the choice of experimental conditions and assessment of experimental
data, but also to guide those who use diffusion or flow tubes or the measured
rate parameters.
Most theoretical descriptions of a diffusion or flow tube have ignored
the source of reactive species -- the electrical discharge -- by the assumption
of 'a specified value for the reactive specie concentration at the discharge-zone
boundary.13-17 Tsu and Boudart were the first to incofporate the discharge zone
in the derivation and the author has elaborated upon this type of theoretical
I
,

description for both diffusion and flow
108
In the present paper, it is shown that these new equations provick the
missing link between the theoretical expressions characterizing the elecwon and
ion concentration distributions within a discharge zone and the expressions
typically used for describing the gas-kinetic reactions of the discharge pro- /
d u c t ~ . ~ % ~ - Suggestions ~ ~ , are given on how averaged first-order bte oonstants
for the production of discharge products can be measured and how they can be used
I
to calculate the steady-state concentrations of neutral species at any point
within the reaction tube. In certain cases, relatively modest equiptitent can be
used to perform such measurements.
I
2
The numbering of equations in the present paper continues a previous
sequence. le
11. RATE EXPRESSIONS
A. Neutral Species
Consider a binary system of atoms and the parent molecules pr-ent within
a long cylindrical tube that has two zones -- a discharge zone and a reactor zone
(Fig. 1). If the walls of the tube are not too active, if there are no threebody
recombination reactions, and if the atom concentration is less than l’$, the
mass- balance equations for the atoms become”
- diffusion convection atom production first-order
atom loss
for the discharge zone and
diffusion convection first-order
atom loss
for the reactor zone. These equations can be
and
rewritten in dimensionless form,
In a previous paper, a total of 36 equations summarized the operation of
a flow OK diffusion tube, with or without a first-order atom loss process in the
reactor zone, for three different discharge- zone boundary condidiona and three
different end-plate boundary conditions.le The solutions to Eq. (7) were mb-
I
1
,‘
I
1

divided into four c.ases:
Case I. Diffusion tube (B = 0) with inactive reactor zone
walls (1/p2 = 0);
Case 11. Diffusion tube (p = 0) with active reactor-zone(
walls (1/p2 p 0);
Case ID, Flow tube (p # 0) with inactive reactor-zone Walls
(1/p2 = 0); and
Case IV. Flow tube (B # 0) with active reactor-zone walls
(Up2 # 0).
The eight most useful solutions are listed in Table 11. The most general
solution for the reactor zone, from which the 35 others can be obtained by
appropriate simplifications, is Eq. (60). The original paper should be consulted
for further information concerning the derivation 'of the mass-balance equations,
the boundary conditions, or the other theoretical solutions. Definitions of
dimensionless groups are given in the Appendix.
TABLE 11. Pertinent Solutions to Equation (7).
L+M
Discharge-zone boundary condition: = 0 at X = -
dX R
Case I. Case 11. Case 111. Case IV.
= 0 at A = -m Eq. (27) Eq. (36) Eq. (27) Eq. (54)
dJr = at X = 0 Eq. (33) Eq. (42) Eq. (51) Eq. (60)
Jr = a262 (27)
--

B. Ions and Electrons
I
Consider a ternary system of electrons and two different positive ions .
present within the discharge zone shown in Fig. 1. If it is assumed that .
Cl+

VI. I+
= 2 [ 1 - p2 ]
4
111
when: I. l/Si2 = 0; 11. the first root of Eq. (115) is dominant; 111. the first
root of Eq. (Us) is dominant; IV. p = 0, T = 0, and I,(%) is large in Eq. (118);
V. 3 = 0; VI. a21+ = 0 and l/6;2 = 0; and VII. p = 0 and 10(1/6$ is large
3 F
in Eq. (120). Carslaw and Jaeger should be consulted for other solutions to
these types of equations.25
An average value for the concentration of the minor ionic specie, q+,
can be obtained by the following integration,
dpdT
Unfortunately, the integral jiIo(anp)dp cannot be evaluated explicitly.
111. CONSEQUENCES
A. Previous Results
For the sake of brevity, a number of topics pertaining to the discharge
zone given previously will not be repeated here. The reader is referred to the
following sub-sections of Sec. VI in reference (19): A. Choice of Steady-State
Systems; B. Complexity of Solutions; C. Velocity and Mass Flux; D. Residence
Time; E. Radial Concentration Gradients; F. Comparison of Rate Processes; I.
Back Diffusion; and M. Surface Reaction vs Atom Production.
B. Measurement of @
1. Comparison of Rate processes
The behavior of atoms in the system shown in Fig. 1 is characterized
by eight different rate processes: convection, radial diffusion to the walls,

'
112
axial diffusion through the discharge one, axial diffusion through the reactor
zone, production, recombination on the end plate, recombination in the discharge
zone, and recombination in the reactor zone. Thus, the interaction between the
rate of production and the other rate processes can be conveniently characterized
by seven dimensionless groups: 11
-
U2
=
RI n
2-
RIL -
rate of atom production in the discharge zone = 'e2
rate of radial diffusion to the wall D12
rate of atom production in the discharge zone
p M
or &R
rate of atom recombination within the discharge zone k: YVl
rate of atom production in the discharge zone
P !Go,%&!
rate of atom recombination within the reactor zone kl yV1
rate of atom production in the discharge zone - W R
rate of atom recombination on the end plate - Y'F1
rate of atom production in the discharge zone E k&
rate of axial convection VZ
rate of atom production in the discharge zone -kdRM
rate of axial diffusion through the discharge zone D12
rate of atom production .in the discharge zone *L I
rate of axial diffusion through the reactor zone. D12
These dimensionless groups can be further condensed to,yield just two
groups relating the overall rates of atom production, transport, and loss:
= rate of atom production = U'
loss rate of atom loss 115' + 116Iz + l/p
-
( 124 I
@loss - rate of atom loss - l/k' + 1/612 + 1/u2 r
-
Otransport
- R/L + RIM + $ + 1
In a steady-state experiment, there must be a competition between two
rate processes -- a source and a sink. If the source and sink are spatially
separated, then a transport process may further complicate the results. In
the present case, the rate of atom production can be measured only if
rod
1, for if the source is much greater than the sink, no concentration
@loss
gradient can exist.
2. Significance of k&
For purposes of discussion, assume that the rate parameter Id represents
the rate of production of atomic hydrogen(c;) via the following ion-electron
recombination reaction,
H$ + e' A H + H
(125)
!
f

I
I
1
I
I
I \
Thus,
kA(c - c;) = 2 = 2kr(H:)(e')
defines the parameter k6. If atomic hydrogen is also produced when water is
present via an ion-molecule reaction,
H20+ + H2
k; changes to
kim
> H,O+ + H
It is clear from the above that kt, is a composite parameter which rarely
can be subdivided into its component parameters from discharge product measurements
alone. Furthermore, as shown in Sec. 11. B., the ion and electron concentrations
are not uni'form throughout the discharge zone, so an average value,
-1
k,, is observed,
Thus, the measurement of absolute yields of discharge products is not a
fruitful technique for studying discharge-zone kinetics. An exception to this
statement occurs when the discharge is initially well characterized and per-
turbations are made on this initial state. For example, in the absence and pre-
sence of water, the ratio of the above rate parameters becomes
+
The quantity kim(H20 )AV can be determined provided that the ratio of
is measurable and the other quantities are known.
3. Interaction Between Discharge and Reactor Zones
values
The activity of the reactor-zone walls and the location of the catalytic
end plate have a measurable influence upon the atom concentration within the
discharge zone and at the discharge-zone boundary. This phenomenon can be
used to advantage for the measurement of absolute or relative values of the
rate constant of atom production k;.
In the absence of reactor-zone sinks, Eqs. (27) and (45) apply,
The atom.production rate constant k: can be determined from Eq. (130) provided
that (a) the amount o f atom recombination on the discharge-zone walls is
known, (b) it remains constant during the measurement process, and (c) it is
the dominant atom loss process. . In an electrode discharge, metal sputtering on

the tube walls occurs continuously and the above conditions are not fulfilled.
With other types of discharges, a period of wall aging may be required.
If an end plate is present, the concentration of atoms at the discharge
boundary (1 = L/R) is given by Eq. (33),
When L/R + 5' >> M,
then
62R L
then J, = a26' as previously. However, if 9
>> E + e',
The rate constant k& can be determined if the recombination coefficient of the
end plate or the diffusion coefficient of the atoms is known.
If only the reactor-zone walls are active, Eq. (36) applies at the
discharge-zone boundary (z = 0),
JI = a262 1 + C R
UM
(E - < 1/4) .
The only new result occurs when ER >> 1, and leads to
CIM
which is no improvement over Eq. (132). Equations (42), (5l), (54), and (60)
lead to progressively more complicated solutions.
(131)
(133)
In the preceding paragraphs we have been preoccupied with the measurement
of g. The equations derived above are equally applicable if k;, is already know
and show how such data can be used to predict steady-state atom concentration
levels in diffusion and flow tubes.
As an example, consider the effect of an end plate on the atom concentration
within the discharge zone. In the absence of the end plate, the concentration
throughout the reaction tube is given by Eq. (130). When the end plate is
present, either Eq. (33) OK Eq. (131) applies for the reactor zone and Eq. (135)
applies for the discharge zone (z assumes negative values only between 0 and -MI.
cosh (g)
sinh M/6R
M .
6R
+ 5' + 6 coth -
With R = 2 cm, Di2 = 1.25 * lo4 cm2/sec (atomic hydrogen at 100 mTorr and 25'C),
L = 10 cm, 51 = 2.5 * lo5 cm/sec (atomic hydrogen), M = 8 cm, vz = 0 cm/sec,
(135)
. ..

Ywalls;s lo-', y'end plate = and kh = 0.100 sec-', the average atom concentration
in the discharge zone is reduced from 14% to 0.19%. It is assumed
here that the end plate has no effect on the electron and ion concentration levels
within the discharge.
k. Mass-Flux Devices
As shown above, the measurement of the absolute value of k,!, requires a
knowledge of either 7, M and D12, or M and 7'. A simpler procedure applicable
to the study of atoms requires the use of a mass-flux device such as an isothermal
calorimeter, which measures the atom flux at a given point within the
reactor zone. l9 Equation (131) becomes
Jls (atoms/cm2 sec) = - D12 "I aZ
=2+2&9L
z=L =O
62R
= 0262
CD~~I
R
Ti+!'+- L 6"R
M
cy2 %2
R
= k&c'
when M >> [k + E) . Thus, the rate of atom production can be determined bv a
mass- flux m\easur&nment provided that the value of M is known and the calori:
meter is the dominant atom sink.
5. Relative Measurements
I
(136)
Provided that certain parameters remain constant and the appropriate
inequality relationships hold, the measurement of relative values of k;, is the
easiest measurement to perform in diffusion or flow tubes. The most important
requirement is that

E2R
M '
$6"

117
V. NOTATION
Del operator (vector)
Infinity
Braces indicating molar concentration of enclosed species
Total molar concentration of gaseous species
I
Molar concentration of atoms in reactor zone
Molar concentration of atoms in discharge zone
Molar concentration of minor positive ionic specie
Molar concentration of major positive ionic specie
Molar concentration of high-energy electrons
Molar concentration of low-energy electrons
Binary diffusion coefficient for atom-molecule system
Ambipolar diffusion coefficient
Exponential
Represents quantity related to specie i
Modified Bessel function
Represents quantity related to specie j
Molar flux of atoms to end plate
First-order rate constant for loss of atoms in reactor zone
First-order rate constant for loss of-atoms in discharge zone
First-order rate constant for production of atoms in discharge zone
Averaged first-order rate constant
Second-order ionization rate constant
Second-order ion-electron recombination rate constant
Second-order ion-molecule reaction rate constant
Length of reactor zone to end plate
Length of discharge zone (or half the length for a symmetrical system)
Radius of reaction tube
Maass-average velocity in z direction (axial direction)
Velocity of an atom
Distance from discharge-reactor zone boundary to end plate (+ direction)

1.
2.
3.
4.
5.
6.
7.
8.
9-
10.
u.
-
12.
13.
14.
15 *
16.
17 *
18.
19.
20.
21.
22.
23.
24.
25 *
26.
27
28.
29.
30.
118
REFERENCES
Whittle, E., DOWS, D. A., and Pimentel, G.C., J. Chem. Phys. 22, 1943 (1954).
Norman, I., and Porter, G., Nature 174, 508 (1954).
Pimentel, G. C., in Fdnnation and Trapping of Free Radicals, (Academic
Press, New York, 1960), p. 69. I
Rice, F. J., and Freamo, M. J., J. Am. Chem. SOC. 73, 5529 (1951).
Rice, F. J., and Freamo, M. J., J. Am. Chem. SOC. 15, 548 (1953).
Goldenberg, H. M., Kleppner, D., and Ramsey, N. F., Phys. Rev. a, 530 (1961).
Berg, H. C., and Kleppner, D,, Rev. Sci. Instr. z, 248 (1962).
Shaw, T. M., in Formation and Trapping of Free Radicals, (Academic Press,
New York, 1960), p. 47.
Dickens, P. G., Linnett, J, W., and Palczewska, W., J. Catalysis &, 140 (1965).
Evenson, K. M., and Burch, D. S., J. Chem. Phys. 9, 2450 (1966).
Fehsenfeld, F. C., Schmeltekopf, A. Lo, and Ferguson, E. E., J. Chem. Phys.
44, 4537 (1966).
Fehsenfeld, F. C., Ferguson, E. E., and Schmeltekopf, A. L., J. Chem. Phys.
g, 1844 (1966).
Wise, H., and Ablow, C. M., J. Chem. Phys. 2, 634 (1958).
Wise, H., and Ablow, C. M., J. Chem. Phys. z, 10 (1963).
Dickens, P. G., Schofield, D., and Walsh, J., Trans. Far. SOC. s, 1338 (1959).
Schulz, W. R.,
Morgan, J. E.,
and Le Roy, D.
and Schiff, H.
J., Can. J. Chem. 40, 2413 (1962).
I., J. Chem. Phys. 2, 2631 (1963).
Tsu, K., and Boudart, M., Can. J. Chem. 2, 1239 (1961).
Rony, P. R., "The Mathematics of Steady-State Diffusion and Flow Tubes.
Factors Affecting Rate Measurements," 2nd Annual Chemical Engineering
Symposium, Stanford, California, 1965. (Copies are available).
Rony, P. R, I. & E. C. (to be submitted).
Biondi, M. A., and Brown, S. C., Phys. Rev. a, 1700 (1949).
Phelps, A. V., and Brown, S. C., Phys. Rev. 86, 102 (1952).
Gray, E. P., and Kerr, D. E., Ann. Phys. lJ, 276 (1959).
McDaniel, E. W., Col1isio.i Phenomena in Ionized Gases (John Wiley & Sons,
Inc., New York, 1-
Carslaw, H. S., and Jaeger, J. C., Conduction of Heat in Solids (Clarendon
Press, Oxford, England, 1959), Section 8.3, p. 2l7.
Jolly, W. L., "The Use of Electric Discharges in Chemical Synthesis,"
Lawrence Radiation Laboratory ~~~~-9591
(1960). (unpublished).
Gould, R. F., Free Radicals in Inorganic Chemistry (her. Chem. SOC.,
Washington, D. C., 1962).
Kondrat'ev, V. N., Chemical Kinetics of Gas Reactions (Addison-Wesley
Publishing Company, Ind., Reading, Massachusetts, 1$4), p. 467.
Kana'an, A. S., and Margrave, J. L., Advances in Inorganic Chemistry and
Radiochemistry 6, (Academic Press, New York, 1964), p. 143.
Hurt, W. B., J. Chem. Phys. 9, 3439 (1966).

L - 2
A = - R
MI2 - L
T = a+- R
v R
@=A
D12
a2
D12
119
APPENDIX
bt2 = 2D12(1
RYTl y'22 (surface phase)
p2- p
klR2
' d
'I+ = c
(gas phase)
(gas phase)
2k R
y = &
V1
1 1
= u2+- 6'2
= n~(2n + 11
an M

120
. Dlschorge zone Reactor zone
1: L Diffusion or
flow tube
Measurement
' device
I I
I
I
I
I
Fig. 1. Schematic drawing of a typical "reaction tube," showing
the location of the discharge zone, the reactor zone, the
discharge-zone boundary (at A = L/R), and the end plate
(at - 0).
Table 1. 'Rate processes occurring within a typical reaction tube.24
Ions and Electrons
in Discharge Zone
PRODUCTION: Ionization
Photoionization
Ion-molecule Reactions
Electron Attachment
Charge Transfer
Metastable Atom Reactions
Loss : Ion-electron Recombination
Ion-ion Recombination
Wall Recombination
Loss to Measurement Device
Electron Detachment
TRANSPORT: Convection
Diffusion
Ambipolar Diffusion
Drift
Cyclotron Resonance
X=O
MU. 3 5708
Neutral Species in
Discharge and Reactor Zones
Ion-electron Recombinat ion
Ion- ion Recombinat ion
Ion-molecule Reactions
Electron Attachment
Wall Recombination of Ions
Gas-kinetic Reactions
Recombination on Walls
Recombination on End Plate
Loss to Measurement Device
Convect ion
Diffusion

c
i
I
I
t
\
121
I
Radiation Chemistry and Electric Discharge Chemistry:
Comparison and Contrast.
Milton Burton and Koichi Funabashi
Department of Chemistry and the Radiation Laboratory"
University of Notre Dame, Notre Dame, Indiana 46556
* The Radiation Laboratory of the University of Notre Dame is
operated under contract with the U. S. Atomic Energy Commission.
This is AEC document number COO-38-511.

122
In radiation chemistry the customary situation is that a charged
particle with energy in the MeV range is slowed down in matter and, in
the course of such deceleration, excites some molecules and ionizes others.
The ultimate fate of the initial charged particle is of little consequence
compared to the many processes which it initiates but that initial particle
is actually responsible for initial excitations and ionizations which amount
to about one-quarter of the total number, of such processes. The corollary 1
statement is that about three-quarters of all the effects observed are
I
attributable to the action of secondary (and, to a very small extent, tertiary) /
charged particles. Roughly, these particles have initial energy, ti, averaging -,
about 75 eV.
The secondary charged particles can cause further ionizations (in
addition to excitations) only so long as they have energy, L, in excess of
the ionization potential, I, of the species which they traverse. A convenient
value for I is about 10 eV. When E becomes < I, the only processes which
can occur significant for radiation chemistry are electronic excitations to
singlet or triplet states. When 6 becomes < 5 eV the major portion of
the excitations are optically disallowed (i. e., mainly triplet). For radiation
chemistry about ? x or 5% of all the phenomena are initiated by particles
4 75
with about that amount of energy.
In electric discharge processes, irrespective of the voltage employed
the electrons start with about zero energy and rarely attain an energy in
excess of 5 eV and much more rarely in excess of the ionization potential.
It has been one of the important problems of electric discharge chemistry
to determine how molecules are excited to chemically active levels and
how ionized species are produced. According to some theoretical views
most of the electrons cause optically forbidden excitations on the first
opportunity and it might appear that chemical effects simply could not be
observed. Nevertheless, they are observed and the fact must be that a
significant number of molecules are excited to sufficiently high states so
that chemistry results. Furthermore, some of them, in number adequate
for maintenance of the discharge, are excited up to ionization potentials.
Thus, in this case, ~i is initially approximately zero and might appear not
to exceed very low energies; e. g., E ~ , corresponding to the lowest triplet
state. Yet, high excitations and ionizations do occur.
Two views may be employed to account for this dilemma. The theory
presented is oversimplified and electrons in significant quantity attain
energies in excess of 7 eV. Alternatively, it was suggested initially by
Magee and Burton' that electric discharge chemistry is characterized by
a series of successive excitations by low-energy electrons in optically
forbidden steps to higher and higher levels with ultimate attainment of
levels adequate for production of a chemical effect - or even of ionization.
The role of thermal energy is somewhat different in the two fields.
In discharge chemistry a local temperature may be raised as high as
1000°C, whereas the usual dosage level used in the basic research of
radiation chemistry is so low that the temperature increase can be of the
order of a few degrees C even if the entire energy input is converted to
thermal energy. This difference is not a trivial one, because thermal
2.
I
<
I
I
l

123 3.
energy and electronic.energy .(in t,he forms of excitation or ioniz,ation) do
not exist as independent sourc-es of energy for, chemical reactions; the !
,of energy is not a ,, .. .
effect arising from a combination of the two.,forms
simple sum of the two effects when each source acts alone ... $'he detail
Of the interaction between thermal and electronic energies is not known.
The qualitative picture, however, is fairly well understood. Namely, the
rate of energy conversion from the electronic to the thermal form (radiationless
transition or energy degradation) increases as the temperature
increases. Thus, the process has the characteristics of an explosive
chain reaction.
The main reason for this difference is attributable to the difference
in the spacial distribution of excited and ionized species initially formed
in each field. In radiation chemistry the distribution is more or less
uniform as a natural consequence of the high penetrating power .of the
impinging radiation and highly random nature of the energy loss processes.
The local inhomogeneities of energy deposition are submicroscopic in size
and cannot be conveniently conceived as local regions of abnormal
temperature. The "hot spot' li idea has little significance in modern
theories of the usual radiation chemistry. On the other hand, in a
discharge tube, electronic excitation and ionization are somewhat confined
to certain parts of the tube (negative glow and positive column). Since
electronic excitation processes are generally accompanied by vibrational
excitation (because of the Franck-Condon.principle), the temperatures of
these regions are.also expected to be higher than those of the rest of the
system. The direct transfer of electron energy to vibrational or
translational motions of molecules can be ignored.
On the other hand, the coexistence of high temperature and high
density of excitation is not really unique to electric-discharge chemistry.
Under extremely high dose rate with certain charged species (particularly
heavy particles such as fission fragments), it is probable that a similar
situation has significance in the radiation chemistry of unusual systems.
A rather obvious feature of radiation chemistry, which cannot be
found in discharge chemistry is the fact that the impinging energy is
theoretically sufficient to excite the molecule to a level where multiplebond
dissociation and multiple ionization are possible. These, effects
have indeed been observed, but their contribution to the total chemical
effect of high-energy radiation turns out to be negligible. The role of
highly excited states or excited ionized states in the physical aspect of
radiation chemistry (energy or charge transfer processes) is far from
being negligible. By contrast, it would appear that lower excited states
alone are the initial products of excitation in electric-discharge chemistry.
In the latter case higher states appear to result from successive excitation
exclusively and, under certain conditions of chemical reactivity, any
presumptive substantial role of such states in the chemistry of discharge
processes can be completely excluded - at least, .on a speculative basis.
Radiation chemistry may be studied in systems of any degree of
aggregation from highly attenuated matter (as in interstellar space) to
liquids or solids under high compression. Studies in electric-discharge

124 4.
chemistry seem limited essentially to gases or to interface.s involving
gaseous systems. In radiation chemistry a major,question. is why, in
spite of the initially high energies involved, the chemical effects are as
specific as observed. In e1ectri.c-discharge chemistry, by contrast, a
very important question is why chemical effects are at all observed and
why they are observed in the high yields reported.
1.
Refe rence s
Milton Burton and John L. Magee, J. Chem. 'Phys., 23, 2194 (1955).
2. F. Dessauer, 2. Physik, 20, 288 (1923).
'
. .I

Inorganic Synthesis with Electric Discharges
William 'L . Jolly I
Department of Chemistry of the
University of California and the
Inorganic Materials Research Division of the
Lawrence Radiation Laboratory
Berkeley, California 94720
Electric discharge reactions which yield thermodynamically
unstable products are of interest to synthetic chemists because such
products are often difficult to prepare by other methods. Many
fascinating compounds of unusual structure have been isolated from
discharge reactions. Although such syntheses usually have low
efficiencies, the general techniques are nevertheless of interest to
chemists who hope to discover new types of compounds.
In this paper,
we shall consider some important types of reactions which can be
effected with the aid of electric discharges, together with some
recent examples of these reactions.
Conversion of Simple Molecules into
Higher Homologs
When simple volatile hydrides are passed through an ozonizertype
discharge tube near atmospheric pressure or through a glow
discharge at low pressure, fairly good yields are obtained of the
higher homologs of the hydrides. For example, both silane and germane
can be converted to their respective higher hydrides; silanes as high
as SieHle and germanes as high as Ge9Hzo are formed. Similarly,
arsine may be converted to diarsine. A wide variety of boron hydrides,
including BgH15, B1&16, B2$16, B&12 and B&2, has been prepared
in electric discharges.
In all these reactions, hydrogen is a by-
product, and there is no tendency for back-reaction of the hydrogen
with the product molecules.
When the volatile halides, SiC14, GeC14, and BC13, are passed
through a glow discharge at low pressure, small yields of the higher
homologs, Si2Cl6, GenCl6 and B2C14, are obtained if provision is made
to separate the products from the by-product chlorine by fractional
condensation. (Chlorine reacts readily with the product molecules
to reform the starting materials.)
improved if a reducing agent which can react with chlorine is included
in the discharge zone or immediately after the discharge zone.
Mercury and copper wool have been found to be very effective reducing
agents for this purpose.
However, the yields are much
In fact, P2C14 can be obtained from a
discharge in H2 + Pcl3 or from a Pc13 discharge followed by copper
wool, whereas no P2C14 has been obtained in the absence of the
reducing agents.

126
Conversion of Mixtures into
More Complicated Molecules
mien mixtures of relatively simple hydrides are passeq through
an electric discharge, higher molecular weight ternary hydrides are
formed. Thus by using appropriate mixtures of SiH4, GeH4, PH3, and
AsH3, it has been possible to prepare, among other things, SiH3GeH3,
SiH3PH2, SiHdsH2, GeHsPH2, and GeHdsH2. By using a silent electric
discharge, which does not cause drastic fragmentation and rearrange-
ment, it is possible, within limits, to tailor-make molecules. Thus
a mixture of SiH3PH2 and SiH4 yields (SiH3)2PH, and a mixture of
Si2E6 and Ph3 yields SizH5PH2.
Passage of diborane-acetylene mixtures through an electric
discharge yields carboranes as the major volatile products, including :
1, 5-C2B3H5, 1,6-C2B4H6, 2, ~ - C E B ~ and H ~ at ~ least six B-methylated
derivatives of these plus C,3-(CH3)2-1,2-C2B3H3. The reaction of
SiH, and (CH3)zO in a silent electric discharge yields mainly a
series of inseparable mixtures, but a fair yield of SizHsCH3 can be
isolated.
A fascinating series of oxygen fluorides (02F2, 032, 04F2,
O5F2 and 06F.7) has been prepared by subjecting mixtures of oxygen
and fluorine to electric discharge at very low temperatures. All
these compounds are very unstable, and, on warming, they decompose
to oxygen and fluorine. Nitrogen trifluoride, AsF5, and F2 when
subjected to a glow discharge at low temperatures yields NF4AsF6.
This material is a relatively stable material which is fairly well
characterized.
The only reported method for the preparation of a xenon compound
that does not involve the use of elementary fluorine or some
fluorinating agent almost as difficult to handle as fluorine, is
the electric discharge synthesis of XeF2 from Xe and CF4. The
carbon-containing by-products of the reaction were not identified.
Controlled Reactions of Atoms and Radicals
with Other Species '
In the discharge methods discussed above, all the reactant
species are passed through an electric discharge. However many
synthetically useful reactions can be carried out in the absence of
a discharge by allowing a stream of atoms or radicals (prepared in
an electric discharge) to impinge on various compounds. For example,
atomic hydrogen is a very reactive species which has been used for
the preparation of hydrides and for carrying out reductions.
promising field of study is the reaction of atomic hydrogen with
aqueous solutions. In basic solutions, the hydrogen atoms yield
electrons, and the method furnishes a convenient method for studying
the reducing effects of the aqueous electron.
A
I

i
\
’.
I
i
t
127
Atoms arid radicals act as electrophilic reagents. The atoms
0, C1, Br, I, and N favor attack at polarizable donor atorps. Thus
atomic nitrogen reacts with divalent sulfur compounds (HzS, Cs2,
OCS, Sa, S2Cl2 and SC12) to yield sulfur-nitrogen compounds, whereas
no sulfur-nitrogen compounds are formed in the reaction of N atoms
with SOe, SOC12, and SO>. The reaction with s2C12 gives fair yields
of the interesting molecule NSC1.
Equipment
Most of the readily-available laboratory equipment for producing
electric discharges is suitable only for relatively small-scale
operation, involving only a gram or two of material. Such equipment,
and the corresponding low yields, is satisfactory for the preparation
of new compounds in amounts satisfactory for identification purposes.
But in order for electric discharge syntheses to be useful for the
preparation of mole quantities of materials, as are the usual synthetic
methods, it will be necessary to have a marked break-through in the
development and marketing of these instruments.

Hydrazine Synthesis in A Silent dlectrical
Dischame.
Thornton J.D., Charlton W.D., SPeddina; P.L.
Department of Chemical bngineering, ,
University of Newcastle-upon-Tyne, Merz
Court, Claremont Road, Newcastle-upon-
Tyne 2. dngland.
IIU!HODlr CT ION.
The eynthesis of hydrazine from ammonia ursing the
eilent electric discharge was firet demonstrated by Beeson
in 1911 (1). Subsequent investigation ehowed that both
the electrical energy yield and percentage conversion
obtained were very low (2,3) and interest in trie procese
waned. Further work was introduced in the early 1950's
enabling substantial improveuents in yields to be obtained
in certain circumstaxcea and a better understanding of the
kinetic mechanisms in the discnarge to emera. Devine and
Burton (4) showed that significant hydrazine fornation only
took place in the positive column of a D.C. discharge. \
Moreover, they found that yields could be substantially
increased if tne atomic hydrogen concentration in the dis-
charge could be reduced by recombination. These general
observations were later confirmed b: Hathsack (5). As a
reeult mechanisms were proposed for hydrazine syntheeie baaed
I on the underlying premise that the hydrazine was fket
formed in the discharge and then degraded by back reactions.
kchi (6) enowed that yielde could be increased by reducing
the residence time of the hydrazine in the diechargo in
agreement with the general premise of hydr85ine dyradation
in the discharge. Subeequent work nas not been at variance
with thie finding (7,83.
Recently, I.C.I. of the U.K. have reported (9) that
removal of product hydrazine in a liquid absorbent give0 a
subetantial increase in yield and an improvement In p ep
centage conversion. Thia ie, of cour~e, a modification of
the general premise of hydrazine degraation in the diecharge
proposed by Devine and Burton (4) and Oucbi (6). Furthermore,
following on the pioneering work of Ouchi (61, and other.
(lo), I.C.I. apparently were able to achieve even better
yields by euitable modification of the waveform characterietice
of the discharge. In order to confirm their clair
and to help clarify the mechanisms taking place in the diecharge,
work wae commenced on this eyetem at the Univereity
of Newcastle-upon-Tyne in 1965. Thie ie a prelimiPary
statement of the reaulte obtained to date.
BXPrnIrt&rn&.
The main ah of thie work was to attempt to hCI=eaEe
hydrazine yielde by reducing the residence tine of the
product in the discharge. A concentric barrier dimcharge
reactor was employed with and without trie uee of a liquid
abeorbent. Reactant flow rate wae increaeed UP to the

+xiip.ni +xmi.ng ca iscity of trie appai,atus after W-LiCii, tile
Ti. ?.I.. e b::s reducer:,'.p;- chaibing the electrode area
r .to, give a .f.urther-.reauc!,ion in product residence
t.irJiei. %'i+ly a d.C. '. paral'lel electroae react& was we'd '.
in ,:w,yicn t.le discharge waveform characteristicg;,werd
altered'. so that only a s:iort activating pulse.; was 'supplidd
, to 'the reactarit in tne eiectrode gap as it passed throqh;,
.. . the ., reaat,or.
,in .a:
flowing gas train. Commercially 'pure amon5d.uas fed ihto
the raeasurinf. section of the ap::aratus via a'reduction
valve, &d- a reigxlating needle .vzlve. The flow rjte: was '.
measured on a rotaJ:ieter wriich had been previsusly calibrated
under operating coilditions by us- a soap film manometer.
Gas te:.lperatures an6. 1.ressures.. also were uie:?.sL.red before the
dischar,,;e reactor. The hycirazine formed in t iie discharge
was absorbed in ethylerie glycol either in situ or in a
separate absorrition trLin. Hydrazine was determined using
the spectrophoiietric method of' W a t t ' and 'cLr?sp (11).
Vacuum control. 'was acuie'ved by a Cartesian manostat located
..
before the 'vacuuh-.puunp.-
The A.C. radio frequency power (1.2meg c/s) to-the
A- and B-type reactors was supplied by a modified C-12
Radyne generator -of law rated output. 3easurement of the
power dissipat,e.d.. in the. aischarge was acnieved by firstly
'.determining t.h,e, power factor directly. on%uitable osc.il10-
'sc3pe. This value in conjunction with the direct
-readings of an X.iG.S. voltmeter (Airmec 314) and a radio '
. frequency ammeter- (.Cambriuge , Unipivot ) . enabled a reasonably
accurate .d,etermination of the power in the discharge .to be
-achieved. The D.C. power for the C-type reactor was
provided by a specially engineered 8KW generator. Measure- .
:merit of the'actual discharge power was made using a
,combination of an oscilloscope trace and the appropriate
meter .r.eadinga.' , . . ,
The vaxioue types of diqcharke reactors used 'in this
work are illustrated schen&%ically in Fig.1. Other essent.ia1
geopetric details of these reactors and their operating data
are given In Table 1. The A-type seriqs of reactors
consisted of a precision silica tube (which acted as the
capacitive barrier) with the high tension electrode attached.
wound the outeide. The inper electrode was a spinning
cylinder so constructed that ,absorbent liqukd could be
eprayed onto the ineide of the tube to flow-down through the
m,ular diecharge gap. Co-current gas and liquid flow was
'. 'reactor 'set brtueen measur mg and analysing sec%,i.ons
: , -;'
\ ..
The apparatus conshod esse:;tially of t-b discharge , : '
employed; : . .
The B-type eerie6 of reactors were of -similar. conetruction
except tllat a central wire electrode vae used and the
absorbent liquid was fed into the incornin@; gas a8 a epray.
Dieperpion of, the liquid vas, achieyed ylt,raeonically using a
vibratory gemrator. The apray was fed into fhe kae atream
through on annular orifice placed at a euffieient dietanbe
, ,, _. ..
,. . '

{;I .
-
. .. . - .., . . .. _- .. ..., ,, : .......- . .:,.: - . , : ,,; - . .. . :;:.:.:.. .:..:.:. ._.__ . ..... .- .:.. ... - :: .
Flg. 1 kh.motlc dlqmm of Factors urd for hyamln
E = E1octrod.r L=UqJd Inlet Q--QI Rorr
S-Sllko Borrlw U=UltramlC Vlbmta
4 6 0 10 12 14 16 18 20 22 24 26 28
Pcwu h i t y KW /ccx103
FI) 2 Hydrotln. y W tor th. cwx~ol rmctor A1 wlth L without WA
!

I
I
t
I,
'\
I
'a
downstream from the discharge to ensure proper dispersal of
spray in the electrode gap. The C-type reactor employed a
pair of rectwar electrodes so set in the gas flow as to
avoid reactant by-passing. "ne use of D.C. power,
necessitated stabilization of the discnarge by means of a
resistive load in the electrical circuit. Provision was
made to admit liquid spray into the inlet gas stream using
the same aerosol generator used in tile type-B reactors.
Table 1.
Reactor units employed.
Type
Reactor Flow Area Discharge Discharge Uilectic
CoGe cm 2
I idth rjarr ie r
cm c1n cm
Tubular 81
Centrifwal A2
Fi3.m ileactor A3
A4
2.203
II
II
n
1.270
0.254
0.127
0.040
2.798
0.560
0.280
0.089
0.15875
II
II
II
rubular
Spray Reactor
B1
62
1.089
II
la270
0.254
1.383
0.277
It
II
Used
Reactor
c1 1.4U 0- 635 0.896 0
"O1Y "hicme ss
gmJLTs:.
Hydrazine yields for the various reactor geometries and
the operating variables employed are given in Figs. 2 tm 6.
From the data presented in rig 2 to 4, it is evident that the
yield varies inversely as an exponential function of the
*lower density at pressures under l0Umm of mercury. The effect
of pressure also follows a Regative exponentid variation
therefore a general equation of the form
Y = a exp (-W-cv)
adequately describes the results. Of course this correlation
of the results does not in any sense give a complete
description of the underlying pnysical chemistry involved in
the synthesis. Ideally,it would have been preferable to
develop separate rate equations in terms of the partial
pressures of tne various components in the overall hydrazine
1 synthesis. This would have necessitated a complete product
analysis and since,in the present case,these data are lacking
the phenomenological approach had to be used. The constants
the equation for tne various reactors acd operating
conditions used were evaluated using the method of least
squares (12). These are tabulated in Table 2. It should be
noted in this Table tnat certLin pressure exponential values
were bracketed. here the relevant data were lacking for
c&cdat&n and ,therefore, the dependence of energy yield upon
preesurefdeactors A2 to A4 and B2 were assumed to be the
eane RS those found experimentally for Reactors Al and B1.

1
\
\
\
,
\ ,
\
\
Pmer hnrity K.W. /ccxiD)

. a b c liqqid
*iea.ctor L.!n/iLWH ~3a-l c c/Kwat t absorbent
Al .l3.14 O.OUB7 73.81 firm
81 6.33 3.02j6 118.74 none
A2 10.70 (O.OU87) 19.43 f iliil
I1
I1
A3 9.56
12-93
11
A4 7.28 3.73
II
B1 12.99 0*0098 73.58 spray
32 9.26 (0.0098) . 13.69 II
B2 8.88 (U.0236) 35.15 none
This is a reasoliable procedure within any one
1,erticular series of reactors such as A1 to 4, but it is
OiQY Eiproximate witii iiiffering reactor series( e.g. Reactor
series A and B)due to dissimilarities in discharge geometries.
JJi,c;uSSI #If.
The i?ost notable overall feature of the results is the
.qrogresuive increase in hydrazine yield obteined by tite use
of a liqu-id absorbent and by pihing tne discharge. The
himest averase. jield was obtained wit!i tile pulsing
technique with or without liqcdd absorbent aid was approxinietely
15 gms if H /KMH. Isolated values as nigi? as 20 @as/
W h were obtaine8 %ut further experimental work is required
before these values can be made reproducible. It is
noteworthy that a yield of 15 pis/KMH is beginning to look
comercially attractive although it must be remenbered that
the fiiL;,.l. product cost will depend to a great extent on the
cost of the product purification process downstream of the .
reactor.
There are several interesting features shown by this
work which bear discussion. From the results set out in
Fig. 2 to 4 it is evident that the hydrazine yield decreases
with increasing discharge power intensity despite a corresponding
increase in the decomposition of ammonia. The only
reasonable explanation for this which 118s been advanced is
that hydrazine is formed in the discharge bj a complex
reaction mechanism and is subsequently decomposed by electron
bombardment or other collision phenomena (4,6 ) .
The use of a liquid absorbent resdts in a decrease in
the slope of the energy yield-power density plot. This is
well illustrated in both Pig. 2 and Fig.4 by comparing the
slope of the plots with and drithout the use of liquid absorbent.
This indicates that hydrazine is being reinoved from the
diecharge by the liquid absorbent instead of being degraded.
If the hydrazine were completely removed so that degradation
did not occur the elope of this plot would be wither
independent of discharge power or possibly positive. The very

.
I operating
fact that the sloGe is still negetive with tne use of liquid
absor.bent nay indicate that a sitnif icant amount of hydrazine
is &till being degraded in the discharge in this case.
Therefore, tiie use of a more efficient absorptlon process would
reasonably be expected to reco.i;er more of the hydragice being
degraded and thus increase yields still further. IuIoreover, the
observed uecreasing effect of pressure on the yield with the
iise of liquid absorbent (shown in E'ig.2)adds weight to this
sv.gge s t ion.
As tile 3ressur.e was decreased beloit 1001!m of mercury the
yield of hydrazine increased steadily while the s!.ope of -$he
,;;-ield-power de:>.sity curve remained co:istaiit for similar reactor
coiiaii;ions such as tiie use of a liquid absorbent.
A reasonable Pxplaiation for this beliaviour is that at the
lower pressures tne eiectrons passing into the discharge ire
less likely to suffer collisions in tne immediate vicinity of
the electrode so that the average enerLy of tiie electrons will
be -high just prior to the required activating collisions.
Therefore, activating collisions are more likely to occur and
yields to iiicrezse corresgondingly. Degradation of product by
electrons is. by' the same arguement, more likely under these
conditions but other product degrading discharge collisional
phenomena e.g. the reaction with hydrogen atoms, are reduced
because of the greater mean free path at the lower operating
pressure. The overall result is that hydrazine yields increase.
Secondary effects become increasingly important at higher
' pressures and it is likely that other variztions in the effect
of.gressure on yield may well occur. One such variation ie
reported to occur at about 5mm pressure where the energy yield
passes through a maximum (8).
Product yields were increased by removing hydrazine in a
liquid absorbent ad 'it was considered that a more efficient
absorption technique should lead to even greater yields. In
order to obtain a more intimate gas liquid disperrknn a spray
reactor was used df the general design Bhown-for Reactors
B1 and 2 in Fig.1. The results (Pig.4) show that yields were
slightly below that of the film reactor. In practice there
were difficulties caused by dissimilarities in the construction
of Reactors A and B which led to radically different discharge
condition8 being obtained.
The film reactor (U) had a more
uniform discharge density because it was of an apnular
construction where the annulus width was small compared to the
reactor diameter. The spray reactor, on the other hand,
employed a central wire electrode and coneequently there was .
a non-uniform field in the discharge gap with a higher .local
diacharge density in the vicinity of the wire electrode.
Because the yield is known to depend on an inverse function
of the power demity,it is only reasonable to expect that the
reactor deeign :B1 would give eomewhat lower yield than the
.Reactor design Al, under. similar operating conditione.
This means that the more intimate gas-liquid contact in the
epray reactor had no measurable effect on the product yield.
-

On the other hand, if hydrazine is formed and degraded
ullifolvly in the activating section of tlie discharge this
result is difficult to explain. It is suggested from this
that,under these conditions, the absorption process is no
lolyer the controlling factor since an increase in, the
overall potential absorption rate,tiirou(jh an increase in the
illterfacial area ,produces no corresponaing rise in Warazine
yields. I~ mag be that liquid surface activation phenomena
are t*&ing place and the increased hydrazine yield which is
observed with the use of an absorbent liquid is caused by
hydrazine formed by otlier mechanisms in which the liquid
surface plays an important role. If this were the case the
liquid film, wnich presents a complete barrier to the discharge,
would naturally give naximal yield and tile spray reactor
would only tend to tllis value in tile limit.
It was found experimentally that a decrease in the
residence time of the recictant in the discharge is accompanied
by an increase in energy yield of ivarazine and a corresponding
fall in percent%e conversion. . The results are detailed in
Blg.5 and appear to be in general agreement with those of
kchi (6). When a liquid,absorbent is used the change of
yield and percent conversion is steady but without the use of
an Crbs'orbent the effect becoines very marked at low residence
times. Ouchi (6) has concluded from his results that soae of
the. hyQatinO formed in the diecharge is preserved from
degradation by rapid physical removal out of the discharge.
This would adequately explain, in general terms, the increase
in energy yield with reduced residence time. The degree of
activation of species in the discharge uust fall as the
reactaqt throughput is increased. This would result in a
decrease in percent conversion as residence time was reduced.
Furthermore, as the concentration of hydrazine governs the
rate of the degrading reaction t!ie smaller. percent conversion
achieved at high flow rates will result in a.higher overall
energy yield being achieved. With the use of a liquid
absorbent soize effect on tile yield may well arise beczuse
of changes in the absorbtion rates due to i;.;creased turbulence
and a decrease in the gas liquid cmtact time.
Another point which .mst be borne in mind is that physical
removal of the product or even,for trlat matter, the use of
the pulsed uischarge tec'mique cannot give better convers,ions
,than that dictated by tne equilibrium concentration of
hydrazine for tne basic reactions t?-king place in the discharge.
Tie use of the liquid absorbent, on the other hand, permits
nigner conversion to be achieved because of tile inability of
tne uiscliarge reackion to reach equilibriun as ludraziae is
being continuously renov4 after it is formed in tile discharge.
The negative sloDe of tne ener$y yield-residence the
plot (Pig. 5), yhen liquid absorbent was used, implies that it is
oossible to increase yields Still furtlier by so:ue otiier
suitable LnoGification of technique.. .
'

_-
of
' .
8. -
7
6,
Residence Time x105 hr
Fig.5 Variation of hydrazine yieldwith flow rate tor
reactor A1 ..
X P
&
e V
d
7
v A A
10 -.-- -- I
-\
'\.
! 1
10 20 30 40
Fig.6 Hydrazinc yield for pulsed D.C. reactor C1
-
A A
0
,

i
137
A'ae rcsu'l'cs, lor tiie ca.

135
Ionic Reactions in Corona Discharges of Atmospheric Gases
M. M. Shahin
Xerox Corporation, Rochester, New York 14603 .
INTRODUCTION
Mass spectrometric studies of electrical discharges have been cdrried out during
the past decade by several workers (1-6) in an attempt to identify the ionic precursors
of a variety of neutral by-products which are formed in these systems. Such studies
are expected to reveal the detailed chemistry of the many reactions that follow the
formation nf primary ions through the impact of energetic electrons on the neutral gas
molecules. and end with the neutralization of the final form of the ion either in the
gas-phase or at the electrodes or the walls of the discharge tube. Thus, the identi-
fication of these ionic reactions will, in many cases, demonstrate the mechanism of
the production of the free-radicals which are the source of the neutral by-products.
Further. such studies would be expected to provide information on the mechanisms of
catalytic or inhibitory effects of trace quantities of certain compounds (e.g. water
vapor) on the formation of these products. Processes such as charge-exchange or ion-
molecule reactions which occur with large cross-sections and have been suspected to
be responsible for such effects will thus be easily identified.
The complexity of electrical discharges however, often makes the interpretation
of mass spectrometric data very difficult. Specifically, the variations of electric
field along and across the discharge tube in certain commonly used discharges (e.g.
glows) often affect the abundance distribution of the various ionic species which are
observed at the mass spectrometer (2,3). This is mainly due to the complex dependence
of the cross-sections of both charge-exchange and ion-molecule reactions on ion
energy as determined by the electric field within the discharge tube. Further com-
plication arises from the formation of ion-sheaths around the walls of the discharge
tube. through ambipolar diffusion. These may strongly influence the sampling of the
discharge by the mass spectrometer. It is therefore essential that the system under
investigation be well understood before the interpretation of the data is attempted.
In this paper, the results of mass Spectrometric investigations on low pressure
positive corona discharges established between two coaxially placed electrodes will
be discussed. This form of discharge has been chosen primarily because its electrical
properties are relatively simple and well understood. Further, the ion-sheath effects
at the point of sampling are minimized because of the low level of ionization in
these systems.
EXPERl MENTAL
Detailed description of the apparatus has already been given (6). Figure 1
shows the schematic diagram of the apparatus. Briefly, a coaxial discharge tube is
operated through a high-voltage d.c. power supply and the current is stabilized
through the use of an external current limiting resistor, R. Electrons moving through
the high field region present only at very close distances to the anode wire will
gain energy from the field and cause ionization of the gas molecules in this region.
Because of the symmetry of the electrodes, therefore, the system closely behaves as
a line source of positive ions which are continuously regenerated. These ions will
then move through the gas, perpendicular to the axis, undergoing various interactions
before reaching the cathode. A smal 1 sampl ing port (IO to 100 microns in diameter)
at the cathode allows a small portion of the ions to escape the discharge tube and be
analyzed by a quadrupole mass spectrometer.
The form of the electric field within the discharge tube is rectangular hyper-
bolic before the discharge is established. As the discharge is formed, however,
the presence of the positive space charge distorts this initial field, causing the

139
latter to attain a constant value for the major distance between the electrodes (7).
Differential pumping of the sampling region and the mass spectrometer allows
operation of the discharge tube between pressures ranging from less than 1 torr to
atmospheric. An electron gun placed before the mass spectrometer serves as a indepen-
dent ionizing source for monitoring the composition of the neutral discharge gas when
no ions are extracted from the discharge tube. For such analysis, calibration curves
made on prepared gas mixtures, showing mass discrimination due to the specific flow
condition are used. A rapid flow of gas is maintained through the discharge tube in
order to avoid appreciable accumulation of neutral by-products of the discharge. For
this purpose a linear flow velocity is maintained such as to completely renew the
gas within the tube in two seconds. In systems where the concentration of the electro-
negative gas (e.9. oxygen) is low, an external source of ionization through the use
of a weak radioactive source (e.g. P0210) is used to stabilize the discharge.
The general problem of mass spectrometric sampling from high pressure sources
has recently been discussed by several workers (8). In systems where condensable
gases are present, the temperature drop following the adiabatic expansion of the gas
through the sampling nozzle may cause condensation if such gaseous components are
present in sufficient concentrations. In the present experiments, the water content
of the gases under investigation has been chosen below 5 x 10-2 mole %. Under these
conditions, it can be demonstrated" that such condensations make negligible contri-
bution to the results.
RESULTS AND DISCUSS IONS
Earlier experiments (6) on corona discharges in air at atmospheric pressure
clearly demonstrated the important role of trace quantities of water vapor in these
systems. In nitrogen, oxygen and their mixtures, where the water content exceeded
4 to 5 x mole %, the dominant ionic species observed at the mass spectrometer
were those corresponding to hydrated proton clusters, (H20),H+. These species were
apparently formed from the interaction of the primary ions with the water molecules
in the system as they passed through the gas to the cathode. In order to determine
the role of various reactions which lead to such clusters, low pressure experiments
were designed with individual gaseous components (e.g. 02, Ng) and their corresponding
mixtures with various quantities of water vapor. These experiments were
expected to reveal the formation of intermediate species and their final conversion
to hydrated protons.
Discharqe in Nitrogen:
Experiments with nitrogen at low pressures, containing various concentrations
of water vapor show several intermediates which are formed through the reactions of
nitrogen ions with water. The results of a typical experiment showing the relative
abundances of various species reaching the mass spectrometer as the pressure of the
discharge is varied, are shown in Fig. 2. In this experiment, it can be observed
that as the pressure is increased, the abundance of the primary ion, N2+, is sharply
reduced while those of the intermediate ions, namely, N4+, N2@ and H20+ rise to a
maximum and then decrease as the tertiary ion H30+ begins to appear. This latter
ion also rises to a maximum at higher pressures as its more highly hydrated forms
appear. Thus the two reactions:
+
N2 + H20 N2H+ + OH, di = -0.6ev
(1)
and
"Calculations based on the maximum number of collisions of an ion with water mole-
cules present up to IOe1 mole % in an expanding gas, assuming cross-sections of
10-13cm2, show negligible contribution to the total collisions that such an ion
and molecule will undergo within the discharge system. These calculations will be
published elsewhere.

GAS OUT
GAS IN
i.ho
QUADRUPOLE
MASS SPECTROMETER
TO 4"PUMP
Fig. I . Schematic diagram of the discharge tube and mass spectrometer;
I, platinum wire anode; 2, cylindrical discharge tube; 3, ext ac ion
electrode; 4, mass spectrometer ionization chamber; 5, focuss "9
electrodes; 6, electron multiplier; 7, etectron-gun assembly.
"
5 IO IS 20
PRESSURE IN TORR
Fig. 2. Variation of the relative abundance of different ions with pre: ure
in.a positive corona discharge in nitrogen containing 2.2 x IO 9 mole %
of water vapor.

N2 + + H20 d H20 19 + N2, c?H = -3.0ev
appear to provide species which can further react with water through reactions (3)
and (4) to form the hydrated proton:
-
N2H+ + H20 H O+ + N2,
3
H~O+ + H o H o+ + OH,
2 3
&i = -3.4ev
AH = -lev
I
(3)
(4 1
The third and the major intermediate, namely N4+ which is formed through reaction of
N2+ with neutral nitrogen molecule (9) apparently also undergoes reactions with ,water
similar to reactions (1) and (2) to form N2H+ and H20+ ion. These reactions are also
expected to be exothermic. The relative abundance of N4+ ion is found to be strongly
dependent on the field strength within the discharge tube. Specifically, the formation
of a space-charge around the corona wire is found to strongly influence the
relative abundance of N4+ and N2+ ions in the system as this space-charge substantially
reduces the field strength within the discharge tube. In the experiments shown in
Fig. 2, the magnitude of E/P, the pressure reduced electric field for the major distance
between the two electrodes, was calculated to be about 7 volt/cm. mm Hg.
TWO other intermediate species have been observed in this system. These ions,
not shown in Fig. 2, are N4H+ and NjH+. They apparently arise as a result of ionmolecule
reactions involving N4+ and N3+ ions in the system with water molecules and
are expected to undergo proton transfer reactions similar to reaction 3,
H~O+.
to yield
Other ions observed were at m/e = 32 and 46. These ions are believed to arise
from trace quantity of oxygen which is inevitably present in this system as a result
of the decomposition of water.
m/e = 46 is NO2+.
The ion of m/e = 32 is apparently 02+ while that of
In a system of pure nitrogen where the water concentration was kept below
5 x 10-3 mole %, under similar discharge conditions as those presented in Fig. 2,
it was observed that the intensity of N4+ ion rapidly increased as the gas pressure
was increased, reaching a plateau after a few mm Hg pressure. This plateau corresponded
to 95% of total ion intensity, with the remaining 5% being mainly due to
N3+ ion. This result supports the earlier reference to the presence of a low E/P
between the two electrodes (9) and the lack of any appreciable ion-sheath near the
cathode.
Discharge in Oxyqen:
Figure 3 shows the pressure dependence of the relative abundance of various
ionic species in a corona discharge in oxygen. This experiment shows several marked
differences from those shown for nitrogen in Fig. 2. The most striking feature is
the presence of large abundances of hydrated forms of the primary ion, namely 02+(H20)1,~
and their dependence on pressure. That the abundances of these ions appear to rise
to a maximum and then decrease at higher pressures with the formation of hydrated protons
strongly suggests their role in the intermediate processes. This evidence together
with the appearance of (H20)2+ ions in the system, and the variation of its abundance
with pressure, i.e. rising to a maximum and then reducing at higher pressures, in-
dicates the operation of.an entirely new mechanism in this system for the formation
of hydrated protons. Moreover, the appearance of m/e = 37, (H20)2H* before the first
member of the series suggests that a reaction other than the hydration of H30+, is
responsible for the fermation of this species. Since the ionization potential (I.P.)
of oxygen (12.07ev) is lower than that of water (I.P. = 12.56ev), a charge-exchange
reaction similar to reaction 2, for nitrogen does not appear probable. Nor is it
llkely that a reaction similar to that of reaction I, to form 02p would occur, as it
appears to be endothermic. It is therefore probable that the hydrated form of the
primary ion, especially the species containing two water molecule may be a precursor
in the formation of the hydrated proton through the reactions:

-
+
( ~ ~ + 0 H ) ~ O ~ ( H~o)~H+ + OH
(H~o)~+ + o2 (H,o)H+ + OH + o2 (7)
Reaction 5 is expected to be exothermic owing to the heat of hydration of H20+ ion.
It may also be possible, though unlikely, that any required energy fqr the reaction
be supplied through the electric field in the discharge.
One other ionic species which may play a role in this system is 0 . However,
since the I.P. of atomic oxygen (13.61ev) is greater than that of oxygen, this species
is not observed in this system as it most probably undergoes charge exchange with
molecular oxygen at pressures used in these experiments. The reaction of OC with O2
to give rise to 0 , if it occurs at all, is not observed in these experiments since
ozone has also a ?:gher I.P. (12.8ev) than that of oxygen and therefore 03+ i s similarly
expected to undergo a charge-exchange reaction with 02.
Three other ions were observed in minor relative abundance in this system. These
ions are OH+, H20+ and H202+. The relative yield of the O p ion is shown in Fig. 3
while the yields of H20+ and H202+, being somewhat smaller, are not indicated. It is
believed that the first two of these ions arise from an ion-molecule reaction and a
charge-exchange process between O+ and water molecule respectively, and are detected
only at higher discharge pressures where such reactions compete with direct charge
transfer of O+ to oxygen molecules. H202+ ion, however, may appear as a result of an
efficient charge transfer process between 02+ ion and the trace quantities .of H202,
apparently formed in the system by the recombination of hydroxyl free radicals, or the
reaction, H02 + H2 H202 + H, hydrogen being supplied through the decomposition
of water.
In a system of pure oxygen, where the waier content was kept below 5 x mole %,
only two ions, 02' and O4+ were observed within the pressure range investigated. The
abundance of O4+ ion rises sharply as the pressure is increased while that of 02+ decreases
and both reach a plateau beyond a pressure of 20 Torr in the discharge tube.
The relative abundance of O4+ in this plateau region was found to be about 62%.
Discharge in Nitrogen Containing 0.12 mole % Oxygen:
In order to determine the role of oxygen in air discharges in the absence of
water vapor, a number of experiments were carried out in nitrogen containin 0.12 mole %
of oxygen. The lower limit of water vapor in these experiments was 5 x IO' 3 mole %.
Fig. 4 shows the results of the relative abundance of various ionic species which are
detected at the cathode as the pressure of the discharge is increased. These experiments
clearly trace the history of the primary ionic species, namely N2" as it is
converted to N4+ and the latter undergoes charge-exchange with the trace quantity of
oxygen present in this system. Thus, beyond a pressure of 20 Torr in the discharge
tube the charge carriers in this system are almost all 02+. The trace amount of N3+
observed in the system is also all removed beyond a pressure of 15 Torr.
It is significant to note that no nitric oxide ion is observed in this system,
indicating that neither the reaction N4+ + O2 d NO+ + NO + N2, nor 0 + + N2
NO+ + NO, occur to any appreciable extent, in spite of the fact that bott these re-
actions are expected to be exothermic.
Experiments in which water was not excluded showed the appearance of many ionic
intermediates of both pure oxygen and nitrogen and their mixtures. All these ions
could be traced and their final conversion to hydrated protons followed.
Discharge in Air:
Experiments in air in the absence of water vapor (less than 5 x mole %)
were carried out in order to determine the importance of ionic species of oxides Of
nitrogen in these systems. The results of these experiments are shown in Figure 5.
+

I
Ii
ZT
PRESSURE IN TORR FOR 02+ ION
IO 15 20
I
25
I
30 35
.08
4-
.3
.2
.I
0
-
-
-
I
5
/
/
/X(H20)2+ '
IO I5 20 25 30
PRESSURE IN TORR
- .06
\
-.04 lI
Fig. 3. Variation of the relative abundance of different ions with pressure
in a positive corona discharge in oxygen containing 2.0 x 10-2 mole %
of water vapor. The pressure scale for 02+ ion is shown on the top
of the figure.
PRESSURE IN TORR
Fig. 4. Variation of the relative abundance of different ions with p
in a positive corona discharge in nitrogen containing 1.2 x
of oxygen and less than 5 x 10-3 mole % of water vapor.
essure
0-1 mole %

144
i
I , , , , .
+
1- u
3c
w
Inu
c c
00
U
.-
U L
c w
wc,
ks!
u-
u-T .- CI
0 .-
w- 3
01
u-u
nJ
- m
w e
L:
w o .
L u*
U
a w
u- >-
0 .- 0
W E
c .- 0 Inn
*- 0 I
u a0
n J -
.-
9
L
m c
X
> .- VI
m
.- m
LA
i
1

As expected from the previous sections, O$4?on appears as the most abundant ion in
this system. Also as expected, 04+ ion appears in the system and as the pressure is
increased, it becomes an important charge carrier and its relative abundance eventually
reaches a plateau. The appearance of the ion 02+(H20), hydrated form of molecular
oxygen ion, in this system, indicates the high affinity of this ion for hydration as
the gas contained less than 5 x 10-3 mole % water in the discharge tube.
Nitric oxide ion was the only oxide of nitrogen found in the system under the
experimental conditions used here. Its relative abundance remainedlconstant as the
discharge pressure was varied between 2 to 40 Torr. The presence of NO+ in these experiments
and its absence in experiments carried out in nitrogen containing 0.1 mole %
of oxygen, indicates that these ions are probably formed through the reaction
N ~ + o2 + NO+ + NO, AH = -4.5ev
+
Assuming that 02' arises eiti4er through charge-exchange with Ng ion or by
direct electron impact on neutral oxygen molecule, one can use the data in these experiments
to obtain a relative ratio of the rate-constants for the charge-exchange
reaction of N2+ with oxygen to that of ion-molecule reaction (8). This ratio is found
to be equal to 8, a value which is lower than the ratio of the published values of
these rate-constants (IO). This indicates that possibly other reactions such as N+ +
02 + NOf 0 and O+ + N2
NO+ + N or others involving neutral atomic species
also contribute to the total yield of NO+ ion.
CONCLUSIONS
Mass spectrometric studies of low pressure positive d.c. corona discharges in
atmospherrc gases containing trace quantities of water vapor show a complex series
of reactions with each component leading to the formation of hydrated protons in the
system. In the case of nitrogen, intermediate species N2* and H20+ are presumably
formed through ion-molecule reaction and charge-exchange of N2+ and N4+ with water
molecules. These species later form the hydrated proton through proton transfer reactions
in subsequent collisions with water molecules. In moist gaseous oxygen, it
appears that the hydrated form of the primary ion 02+(H20)2, plays an important role
in the conversion of the charge carriers to hydrated protons. It is suggested that
this transformation may occur through the formation of the intermediate (H20)2+ which
has been found in this system. In experiments where water vapor is excluded from the
system, a concentration of 1.2~ 10-1 mole % oxygen can transform, through chargeexchange
reactions, all ionic species of nitrogen to 02+ at a discharge pressure of
20 Torr, and that ion-molecule reactions leading to the formation of oxides of
nitrogen are by far less probable within the pressure range investigated.
ACKNOWLEDGMENTS
The author is indebted to Dr. W. Roth for his invaluable comnents during the
course of this work and to A. Friske for his assistance in experimental work.
References
Knewstubb, P. F., Tickner, A. W., J. Chem. Phys. 36, 684 (1962).
Knewstubb, P. F., Tickner, A. W., J. Chem. Phys. 36, 674 (1962).
Knewstubb, P. F., Tickner, A. W., J. Chem. Phys. 11, 294 (1962).
Shahin, M. M., J. Chem. Phys. 3, I798 (1965).
Shahin, M. M., Advances in Chemistry Series, No., 58, Ion Molecule Reactions in the
Gas Phase, p. 315 (1966).
Shahin, M. M., J. Chem. Phys. 45, 2600 (1966).
vonEngel, Ionized Gases (Oxford University Press, London, 1965).
Green, F. T., MiIn, T. A., J. Chem. Phys. 39, 3150 (1963); Leckenby, R. E.,
Robbins, E. J., Treval ion, P. A., Proc. Roy. SOC. (London) G, 409 (1964);
Anderson, J. B., Anders, R. P., Fenn, J. B., Advances in Atomic and Molecular
Physics, p. 345; edited by D. R. Bates, Academic Press, New York 1965.
Varney, R. N., J. Chem. Phys. 2, I314 (1959).
Ferguson, E. E., Fehsenfeld F. C., Golddan, P. D., Schmeltekoff, A. L.,
J. Ceophys. Res. 70, 4323 (1965).

146
SYNTHESIS OF ORGANIC COIGhTDS BY TKE ACTION OF ELECTRIC
DISCHARGES IN SIMULATED PRIMITIVE ATMOSPHERES
C. Ponnamperuma, F. Woeller, J. Flores, M. Romiez, and W. Allen
Exobiology Division
National Aeronautics and Space Administration
Ames Research Center
Moffett Field, California
In the study of chemical evolution we are interested in the path by which
molecules of biological significance could have been formed on the primitive
earth in the absence of life. It is generally accepted that the primitive
atmosphere of the earth consisted mainly of methane, a-monia, and water. Various
forms of energy such as ultraviolet light frm the sun, electrical discharges,
heat, and ionizing radiation acting on this atmosphere must have given rise to
a wide variety 3f organic substances.
Table I gives a summary of the sources of energy 03 the earth's surface
today. (1) It is probable, therefore, that solar energy must have made the
principal contribution to the synthesis of organic compounds in primordial
times. Next in importance are electric discharges, such as lightning and corona
discharges from pointed objects. These occur closer to the earth's surface and,
hence, would have more effectively transferred the organic matter synthesized
toths primitive oceans. This paper describes attempts to simulate some of the
reactions which may have taken place on the prebiotic earth, through the action
of electric discharges.
While extensive work has been done on the effect of electric discharges on
various organic molecules, relatively few experiments have been performed to
elucidate its role in che.mica1 evolution. Some of the earliest such investiga-
tions were carried out by the chemist Haber.
Beutner, in his book entitled "Life's
Beginning on the Earth," recalls how Haber performed numerous experiments in which
electrical discharges were sent through carbon containing gases like methane,
carbon dioxide, etc., with the aim of obtaining sugars. (2) Although traces
of some sugars were formed, a large number of various other substances were
also synthesized. Haber thus came to the conclusion that by means of electrical
discharges through carbon containing gases, "practically any substance known to
organic chemistry can be found."
Perhaps the most celebrated experiment in this field was performed by Stanley
Miller in Urey's laboratory in 1953. (3) Miller submitted a mixture of methane,
ammonia, and water in the presence of hydrogen to electrical discharges from tesla
coils. A large number of organic compounds were formed. Among these, four amino
acids were identified. Miller postulated two alternative possibilities for the
mechanism of synthesis of amino acids. According to the first, aldehydes and
hydrogen cyanide are synthesized in the gas phase by the spark. These aldehydes
and hydrogen cyanide react in the aqueous phase to give amino and hydroxynitriles.
These nitriles are, in turn, hydrolyzed to amino and hydroxy acids. The mechanism
is essentially a Strecker synthesis. A secmd suggestion made was tkat the amino
and hydroxy acids were synthesized in tk gas phase by ions ad radicals produced
in the electrical discharge. MiUer's subsequent work has shown that the first
mechanism is the one most likely to have produced the amino acids. (4) The rate
of production of aldehydes and hydrcgen cyanide by the spark and the rate of
hydrolysis of the aminonitriles were sufficient to account for the total yield
of amino acids.
I

\
. 147
In Our experimental work, we have endeavored to study the mechanism by which
various molecules of biological interest could have been formed by the action of
electrical discharges. A series of experiments was outlined in which the starting
b materials were vaxied. Four different classes of experiments have been performed;
with methane; with methane and ammonia; with methane, ammonia, and water; and with
methane and water.
1 -
I
The effect of a semi-corona discharge, a low intensity arc discharge, and a
, high intensity arc discharge on gaseous methane was first investigated. The current
through the discharge was measured by the voltage drop across a resistor with the
( Cell. For the semi-corona discharge, the cell current was 0.4 m. amp, 0.5 m. amp
, for the low arc, and 10 m. amp for the high arc discharge. Gas chromatography and
mass spectrometry were used for the analysis of the end products. Comparative ret
sults of the analysis of hydrocarbons up to C5 are shown in Table 11. In the semicmona
most of the methane remained unreacted after a 24-hour discharge. Some ethane
and propane were formed. There were small amounts of ethene, propene,-and substituted
paraffins. In the case of the low and high arcs, ethylene and acetylene were also
> present.
The analysis of the hydrocarbons fram Cg - Cg reveals that the semi-corona
gave unsaturated substances while the arc discharge gave rise to aromatic compounds.
i The semi-corona cell'yielded a colorless distillate, the gas chromatogram of which
was poorly resolved. The high intensity arc gave a yellow fluid, the chromatogram
\ of which had well spaced peaks. Benzene was the most abundant with toluerenext in
order of magnitude. The peaks from the semi-corona chromatogram were identified by
, the use of mass spectrometry as: 2,2-dimethyl butane, 2-methyl pentane, +methyl
I pentane, 2,b-dimethyl hexane, 3,4-dimethyl hexane. In figure 1 the chromatogram
:I of the semi-corona discharge products (low) has been superinposed on that from the
~ high intensity arc (high). The results presented here show that the character of
compounds in the range 3f interest appears to be determined by the type of discharge
1 more than by any factor.
(5)
1 We have also examined the composition of the hydrocarbons above C9 in the products
of the semi-corona discharge. The gas chromatogram is very unresolved (Fig. 2).
No normals 31' branched-chain isoproprenoid hydrocarbons were identified. Analysis
\ of the mixture by mass spectrometry shows that the compounds are possibly cyclic
1 in structure. (6)
The effect of an arc discharge on anhydrous methane and ammonia was next inves-
1
tigated for two reasons. Firstly, such a study would help us to understand the
' pathways by which some organic compounds such as amino acids can be synthesized.
Secondly, reactions of this type would simulate, to some extent, conditions which
may exist on the planet Jupiter.
'' In this investigation, we have used reaction vessels of about a liter in
' vglume containing an equimolar mixture anhydrous methane and ammonia up to a pres-
),sure of 0.5. The electrdes consisted of gold wires about 1 cm apart. A typical
reaction lasted for about 1.5 hours. The current passing through the- system was
about 0.5 mA. The end products consisted of: (1) gases, (2) a colorless distillate,
,'\and (3) a ruby colored residue.
1
1 distillate.
In the present study, our attention was prinarily directed to the colorless
The volatile products were vacuum distilled into a U-trap at -78Oc and
analyzed by gas chromatography (Fig. 3). The fractions corresponding to each peak
The GLC retention time,
'' were collected for subsequent mass spectrometric amlysis.
1 the mss spectrometric fragmentation pattern, and the NMR spectrum established the
k
\
.)
Y,
t
I
1
I
i

mi5
a-
dz
CI
n
I
a
I
n
+4 - 0
-I- k
7
0 -
v)
Q)
t
3
c
.-
E
a
u
I
\o
u
V
k
Ld
0
Ld
2
0
0
V
I
Ti
$
d
hD
.r-
F

i
I
/ '
I
",
m
a 2
k
F4
a
I
cu

0
z
cu
I
I
I

151
identity of each cf the fractions separated by gas chromatography. Ammonium cyanide,
methyl cyanide, ethyl cyanide, a-minoacetonitrile and its C-methyl and N-methyl
., homologues were identified. The u-aminonitriles on hydrolysis give rise to a-amino
acids. These nitriles my prcvide a reasonable pathway for the origin of amino
' acids under prebiological corditions.
In sane cf our discharge experiments, we turned to the question o the origin
of lnonocarboxylic acids under prebiotic conditions. If we assume that pre-existing
abiotically synthesized fatty acids were necessary for the functioning of selective
menbranes, sone rnechanisn must have existed for their formation. The reaction be-
> tween methane and ammonia appears to provide such a pathway. When a mixture of
methane and water exposed to a semi-corona discharge and the end products examined
l after saponification, the monocarboxylic acids from C2 - CQ were identified. (7)
\
The volatile acids C1 - C8 were examined as their free acids by gas chromatography.
A resulting chromatogram is illustrated in Figure 4. The. individual peaks
' were then trapped and their identity confirmed by mass spectrometry. 'Acids contain-
\ ing seven or more carbon atoms were analyzed as methyl esters. The methyl esters,
' after chromatography, were exarnined by mass spectrometry. Of eleven major peaks
obtained by gas chromatography (Fig. 5) only one appears as the normal methyl ester.
Presumably, the renaining peaks represented branched-chain isomers.
~
'
While it is clear that in the case of the longer chain fatty acids several
isoners have been produced, only a few of the innumerable possible compounds are
realized. A preferential synthesis of some type appears to be favored. Theoretically,
the branching of carbon chains, which is favored in free radical reactions, I
may be repressed by steric restrictions when the lengthening carbon chains are
\ absorbed on monolayers. An attempt to favor the for-mation of straight chain acids
by placing the aqueous phase in close proximity with the discharge zone did not
produce any change in our results.
1
In the study of prebiotic organic synthesis, perhaps the most relevent experinents
involve the use of all the main constituents of the presumed primitive earth
' atmosphere. We have therefore exposed a mixture of methane, ammonia, and water to
a discharge from tesla coils simulating lightning on the primitive earth. At the end
' 3f a 24-hour discharge, the gas phase analysis has shown that over 90$ of the start-
'1 ing methane has been converted into organic ccmpounds. Of this, about 45$ is found
in the water fraction. 188 of the water soluble material is in the form of cyanide.
- The formstion of cyanide in this reation is significant in the light of the multiple
role played by hydrogen cyanide in organic synthesis. (8)
The analysis of the end products of this reaction by pper chromatography re-
, veals that a large number of organic compounds were formed but none of these cor-
responded to the commonly occurring amino acids.
A certain amount of material
\ appeared at the origin. However, when the reaction products were hydrolyzed with
6N HC1 for 24 hours and then analyzed, a large number of amino acids were formed
\ (Fig. 6). Among those identified me nine which are commonly found in biological
materials: glycine, alanine, aspartic, glutamic, threonine, serine, isoleucine,
leucine, and phenylalanine. The results obtained by ion exchange analysis were
2 confirmed by gas chromatography. The evidence, thus, points to the fact that the
:lamino acids were already polymerized in the solution of end products. Separation
1 by the use of a biogel-P column gave us a fraction having a molecular weight in the
range 186 to about 2,000 and whose 1-dimethylaminonapthalene -5 sulphonyl chloride
' (DNS) derivative showed a single band on electrophoresis. When this fraction was
hydrolyzed, the aMno acids aspartic, serine, glutamic, glycine, and alanine were
' obtained.
1,
)
\
1:
T

I-
I - t
d
m
a
.rl
Y
x
c,
t-'
d
k

0
0
AI
r
0
z
a
rc)
I
Z
c.
*
I
0
z
0
W
(3
E
a
I
0
m
5
z
0
K
LL
0
Z
z
-
a
3NllVA
-
N- u? AI- . . 0
*a -0
C
cu-
3N13031
((IlJVdNVlS
lVNL131NI)
3NINOIH13H I
3NllSA3
-a -0
. .
AI-
i
i
f

,
155
This result is significant in the context of chemical evolution. It has
generally been thought, that amino acids had first to be synthesized and then
cmdensed together into a polymer. The synthesis of a polypeptide in an electric
discharge experiment reveals that such a sequence of r-actions may not have been
necessary. If a suitable condensation agent is present the polymer appears to be
formed as sson as the acids are s-ynthesized. In our case, the condensation agent
is probably hydr-gen cyanide.
The presence of 18$ hydrogen cyanide in the reaction
mixture combined with the fact that in previous experiments we have beeh able to
condense bases and sugars with cyanide support this hypothesis.
The different experlments that have been described so far reveal that impor-
1 tant biclogical molecules can be synthesized by the use of a form of energy which
existed -n the primitive earth. These conditions may be considered to be genuinely
abiotic since the materials used are the constituents of the presumed prinitive
, earth atmosphere, the conditions are aqueous, and the form of enera is on2 that
is likely to have sccurred on the earth before the appearance of life.
\
References
> 1. S. L. Miller and H. C. Urey, Science ', a, 245 (1959)
2. R. Beutner, Life's Beginning on The Earth, Williams and Wilkins, Baltimore, 1938
3. S. L. Miller, J. Am. Chem. SOC. a, 2351 (1955)
4. S. L. Miller, Bichim. et Biophys. Acta 23, 480 (1957)
5. C. Ponnqperuva and F. Woeller, Nature m3, 272 (1964)
6. C. Ponnamperuma and K. Pering, Nature 2- 979 (1966)
7. W. A. Allen and C. Ponnamperuma, Currents in Modern Biology (in press) (1967) I
8. R. Sanchez, J. Ferris, and L. E. Orgel, Science 153, 72 (1966)
. TAELE I
SOURCE -
Ultraviolet light ( 2500 i)
Electric discharges
Radio activity
Volcanoes
TABLF: I1
High Arc
Total hours of current flow 1- 5
Electrode voltage
Cell current (m. amp)
$ L ss CHq/h
End prsducts ($)
Ethane
Propane
Ethene
Pr=pene
Acetylene
1400
4.0
6.6
32
2
32
27
--
ENERGY
(in cal em+ p-1)
570
4
0.8
0.13
Low Arc
40
2500
0.5
1.1
20
5.2
2.4
1.9
42.5
Semi -Corona
48
9400
0.3
0.6
58. a
36.8
1.5
0.6
0.0

156
The Dissociation of Metal Halides in Electrical Discharges
F o KO MCTaggart
Mdsion of Mineral Chemistry, C.S.I.R.0. Box 124, Port
Melbourne, Victoria, Australia.
I
Introduction.
in electrical discharges in dissociating gae molecules is well known,
and is the basis of the chemical reactions that a H observed in plasnrs
systems. Since gases such a8 02, H2, B2dC1 9 88 well a8 vapours of
organic and other compounds have been diesociazed to atoms ad/m
radicals, it appears Logical to suppose that many other molecules would
behave similarly. This paper describes the diesociation of the halides
of the Croup I, I1 and 111, and rare earth metals.
&m-imental. Ideally, the halides should be introduced Into
the discharge tube in the form of vapoure at suitable pressures. Due
to the wide range of volatilities occurring in the above compounds and
to other erperimental difficulties it proved more convenient to we
carrier gmee, and to volatilize the halides from small boats contained
wlthin the reaction tube by heating these boats to suitable temperatures.
It was usually possible to make use of the heat generated in the plasma
itself for this purpose. Samples were placed upstream from,or within
the central plasma region,ln positione where heat sufficient to give the
desired rate of volatilization was produced.
The effect of the energetic electrons produced
Since heating of solib
due to atomic recombination ie largeu a surface effect it was also
possible to verg the sublimation rate of samples by changing the Lmrfece
area exposed to the plasma.
Both inert gases (in particular He and H ) and hydrogen have been
used and certain reactions which are discussed befow appear to involve E
atoms aa well a dissociative effects. These gases were of coomerciel
purity, the He and 5 being purified by passing them over zirconim ponder
heated to 800°C; although no essential differences between the cylinder
gases and the purified gases were observed. Pressures of the carrier
gaaes were varied between 0.5 ad 2.3 mm. Halides wed were of A.B. grade,
but in certain cases impure compounds were deliberately used in order to
determine the effects of impurities on the dissociation and on the purity
of the metal obtained.
Ertenslve use was made of a 2450 mC magnetron generator (Mullard
m2/205A) which although capable of an output in exce88 of 2 M continuow
wave power, was seldom operated at inputs higher than 600 watts. B&P
frequency energy from the magnetron was fed via a 50 ohm air dielectric
coaxial line, to a section of waveguide operated ea a resonant cavity by
means of a sliding plunger inserted into the open end. The discharge
tube passed tranevereely through the guide at a point of m a r i m electrostatic
field. h e investigatlons, Involving the chlorides of Li, Be,
and a, which dissociate very readily, were made using au 9-f" generator
consintlng of a 4-125A vacuum tube at a frequency of 30 mC, the power of
which could be varied from close to zero to about 400 watts. In these

Compound Bond
he rgy
Li I 82
UBr 102
LICl 115
Lilr 138
157
cases the reaction tube passed through the "tank coil" of the output
circuit . Conventional methods of analysis were used to determine 8/ the
SmOunt of metal deposited, b/ the undissociated halide sublimed onto the
reaction tube, and c/ the halogen or hydrogen halide collected in the
liquid sir trap which followed the reaction tube in the flow1 syatem.
- Results.
From all the halides studied It was poesible to
separate metals, often in good yields.
1
The lithim -UP cornpoonds dissociated in both inert carrier gmes and
in H to gLve metals d the relative reaction rates were detedned for
the fodides, bromides, chlorides and fluorides of Li, Ha, K end Cs. Son18
Of these are ehown in Table 1. It rill be noted that for any one metal
- Table 1
Relative Reaction Ratee for the Hdidee of LI and la
Re10 React. Compound Bond Bel, Bead.
Bate -rgJ Bate
232.5
101.5
66.5
19.0
BaI
BaBr
HaCl
BaF
72
89
99
108
26 e0 15.0
IOIO
5 *o
I

Table 2
Dissociation of NaCl in H2 at 1.0 mn for Various Input Powers
t - -
4 *O I
0.240 0 280
I
I
F'igure 2 shows the effect of carrier gas pressure at constant
power. A marimum in metal production occurred at about 1.5 mm pressure.
For H2 this pressure coincides closely with the highest concentration of
H atoms, and is probably related to the concentration of electrons having
sufficient energy to cause dissociation. The reaction rates appeared to
be substantially independent of the nature of the carrier gas,
The fact that considerable quantities of highly reactive metals
are deposited in atmospheres of even more highly reactive gases, namely
halogen atoms, suggests that these halogen species may be in the form of
negatively charged ions.
Such ions, having a complete outer shell of
eight electrons, would be non-reactive chemically, and if neutralization
of their charges is delayed until they are swept clear of the metal
deposit, an explanation of the apparent absence of appreciable back
reaction would be afforded. Mass spectroscopic studies are being made
on the nature of the species in these discharges to elucidate this matter.
The &OUR I1 compounds. The majority of the halides of Be, ldg, Ca, Sr
and Ba were investigated. Rates of dissociation did not appear to vary
as widely as those of Group I. As optimum conditior.8 were approached
for each compound, all the halide vapourized was dissociated and deposits
were formed on the W d h of the reriction tube in which metals and dihalides
were founO in equi-olar proportions. This points strongly to the
formation of unstable monohalide molecules in the discharge, which dis-
'proportionate to yield the observed products. Again the dissociations
were not dependent on the carrier gas employed. BeC12, one of the more
volatile halides in this group, end one of the most easily dissociated of
all the compounds studied, may be broken down in the lower frequency
apparatus mentioned previously-
Metals could be separated from the dihalidss in the deposits by
means of suitable solvents for the letter, or in some cases by vacuupl
sublimation of the dihalide.

,
”..’
3or
25 -
5 20 -
s
3
w
Y 15-
<
z’
1
f 10-
I
0 8.0
SODIUM CHLORIDE SUBLIMATION RATE (MMOLE/HR) AT Po0 W

160
Group I11 compounds. No in estigations have been made on the compounds
of boron. Ywkovskii t& &.' reported in 1958 that BC1 was reduced to
elemental boron by H atoms, and the present work on alum?nium halides
supports this claim.
When inert csrrier gases were used there was a very limited
deposition of metal from Al and Sc halides amounting to onu a few
per cent of the vapour passed through the discharge, but when hydmgen
wee employed much higher yields of metals were obtained. It would
appear, therefore, that at least one step in the decomposition involves
a H atom reaction, and the process may well be, for AlCl :- initial
dissociation of UC~ to uc13 : H atom reduction of UC~, to UCI :
disproportionation 03 fic1 to U and fic13. A maas spectromet+c
examination of the species present in the discharge revealed AN1 In
readily detectable quantities but no AlCl was observed. Yields of up
to 55% Al metal have been obtained from Ab1 in a single pees through
3
a 2450 mC plasma, and if a second discharge was produced downstreem
more Al deposited. The rate at which AlCl decomposes, with
comparatively low power, and with ita intenae blue plama makes this
reaction one of the most spectacular of those studied. The relatively
involatile AlF3 also dissociated quite readily to glve the metal,
although it is not easily handled in the experimental 8pparatUe described.
The chloride and fluoride of ecandiun behaved in a similar manner to the
corresponding aluminium compounds.
NCl also dissociated readily in the lower frequency apparatus.
It should be Lntioned here that the chief difference between the wavesuide
and the coil-coupled apparatus appear8 to be that in the formar
the electrons are accelerated to higher energlea since the electrostatic
field is concentrated between the resonator walls where a high potential
gradient exists in a direction axial to the discharge tube. By
comparison the E field in a coil is much more randomly distributed.
Rare Earth Comounda. We have not as yet had an opportunity of studying
these halides systematically, but preliminary erperiments have shorn
that they behave in a menner similar to AlCl That is, there is only
limited dissociation to metal in an inert c&ier gas, but with hydroem
satlafactory yields of metals are obtained. Thae Ce and Le result
from &F3, CeC13, LaB and LaC13. If, for example, chlorides contsirr
ing appreciable quautjties of oxychloride are used for the dirsooiation
this inpurity is left in the sample boat and a highly pure metal is
deposited. Such reactions may therefore prove of use in the preparation
of certain metalm.
2. Markovakii, L. V., Lvova, V. I., Kondrashev, Y. D.,
PO Bora i Ego Sveedin, 36 (1958).
Ber. Tr. KO&.
1
i

I
161
THE GLOW DISCHARGE DEPOSITION OF BORON
A. E Hultquist and M. E. Sibert
Materials Sciences Laboratory
Lockheed Palo Alto Research Laboratory
Palo Alto, California
INTRODUCTION
Many ncw materials concepts are evolving out of aerospace materials technology. One of
' the most promising of these is that of boron filament-reinforced composite structures.
Incorporation of continuous boron filament in a suitable matrix could yield a structure with
' the strength of high-strength steel, the rigidity of beryllium, and the density of magnesium.
A large-scale development effort is currently concerned with fabrication of such structures.
, Once this has been achieved, large amounts of filament must be made available at reason-
able cost to make large-scale usage practical. Present techniques for filament production
' are based on scaled-up laboratory chemical vapor deposition procedures, and major tech-
nological development is required to result in large-scale production. As a result, many
i other new and novel techniques for filament preparation have been investigated with the
, intent of developing alternate approaches.
This study is concerned with investigation of one such approach, that of deposition in an
electrodeless glow discharge. Basically a high-voltage, low-amperage, high-frequency rf
currcnl is imposed across a boron-containing gas, resulting in a high degree of ionization/
' activation of the ions present. Boron then deposits out in elemental form on all surfaces
within the glow discharge area. Use of proper deposition conditions confines the glow
largcly to the filament substrate. By passage of the filamentary substrate continuously
' through the discharge, the approach is potentially capable of high rate filament formation
with excellent uniformity and reliability. Variation of current input can result in deposi-
L. tion at any point from room temperature up to 800- 1000°C.
Both metallic and non-
, metallic substrates can be employed. Deposition is achieved at low pressures of the order
of a few mm.
reaction of boron trichloride and hydrogen in a glow discharge. However, efforts to dupli-
cate this work indicate that the conditione employed lead to effects more like an arc than
a glow discharge. Related, although distinctly thermally rather than electrically initiated
I

162
EXPERIMENTAL WORK
Apparatus. The basic apparatus utilized for parametric studies employing a fixed substrate
is shown in Fig. 1. The reaction vessel is a simple pyrex tube with the substrate filament
suspended down the central axis and suitably sealed at each end. The reactant gases are
admitted at one end of the tube and exit at the opposite end. Electrodes are simple con-
centric copper strips which may be moved along the tube for optimum positioning. The
glow discharge then forms between these electrodes. Both horizontal and vertical reactor
configurations have been employed; the vertical system is preferred since dubstrate align-
ment is better controlled. The vertical arrangement is also more amenable ,to moving
filament systems for continuous filament preparation. Both air and water cooled reaction
vessels have been evaluated. Water cooling was originally instituted to minimize sidewall
deposition; later work demonstrated that this could be accomplished by regulation of gas
composition and current input so as to confine the glow area to the central area of the
reaction tube. A three compartment chamber was used so that these experiments could be
performed before the system was opened for examination.
The balance of the system is largely a gas flow system. Hydrogen is passed through a
catalytic cartridge and a liquid nitrogen trap for removal of oxygen and water impurities.
Boron trichloride is fed into the hydrogen stream. Both lines have appropriate flowmeter
devices. The boron trichloride is controlled at the cylinder valve; satisfactory control
with respect to potential condensation is obtained by bleeding the gas into a section of tubing
at operating pressures and heated to 35°C. Hydrogen flow is regulated by a stainless steel
steel needle valve just downstream from the flowmeter. The combined hydrogen-boron ,
trichloride stream then passes into the reaction chamber. Exit gases from the reactor
pass successively through a trap for condensation of unused boron trichloride, past a
manometer, through a Cartesian manostat, and thence to a mechanical vacuum pump.
The power supply is a Lockheed-designed 3 Mc oscillator capable of about 30 W output;
however, only a portion of this output is available as rf power, about 20 W at 1400-1500 V.
The first 20 turns of the large coil (Fig. l), and the capacitance connected across them,
act as the tank. The tank is a resonance combination wherein the energy is alternately
stored in the condensers and in the field of the coil, with interchange occurring at a fre-
quency determined by the coil inductance and capacitance of the condensers. A variable
condenser included in the tank allows its frequency to be matched to that of the secondary.
When such matching occurs, the coils are efficiently coupled, and rf power is drawn from
the secondary. This power is fed into the primary by a 6146A transmitter tube activated
at proper frequency by feedback from the tank. This is done through a condenser and a
resistor wound with a coil, serving as a low-Q resonance circuit to suppress high frequency
parasitic oscillation. A 47K resistor allows the grid to be self-biased. An rf choke pre-
vents loss of power into the backup power supply. TWO 1OK resistors and a 7500 fl resis-
tors serve as a voltage divider for the screen grid.
Materials. The boron trichloride used was electronic grade supplied by American Potash
and Chemical Co. Hydrogen was a high purity laboratory grade (BB-H-886) supplied by
Air Reduction Co. Quartz monofilament utilized as a substrate for most of the work w88
obtained from the General Electric Co. Glass filaments were drawn from pyrex glass rods
by the LMSC glass shop.
General Procedure. Experiments were started by evacuating the assembled apparatus to
1 mm Hg pressure. The leak rate was determined by turning off the pump and noting the
increase - in Dressure over extended time periods. A leak rate of less than 0.05 mm Hg/min
is necessary to avoid boron oxide formation. When the system was determined to be essentially
leak free, a flow of hydrogen was introduced and allowed to sweep through the apparatus
for 15 to 30 min. The desired gas pressures and flow rates of hydrogen and boron
trichloride were established, the power supply for the oscillator turned on, and the flow Was I initiated by adjusting the variable capacitor. At the end of the experiment, the flow discharge
and the boron trichloride flow were turned off and hydrogen allowed to sweep through the

Fig. 1 Process Apparatus
-__l.ll__ I _I_ -
____
Fig. 2 Boron on Glass, Longitudinal Section (3000~)
Boron
B2°3
oxygen
Enriched
or
Oriented
Material

164
apparatus for 10 min to remove unreacted boron trichloride. The pump was turned off and
the apparatus returned to ambient pressure using hydrogen. The apparatus was then dis-
mounted and the sample removed.
RESULTS
Preliminary Experiments. Initial experiments were made to observe operating character-
istics while cxciting hydrogen gas in a glow discharge. Glow discharges were initiated in
hydrogen at pressures of 6, 15, 36, 40, 60, and 81 mm Hg and at flow rates of about 40 to
100 cc/min. Electrode separations of 0.75, 1.0, 2.0, 3.5, and 4.5 cm were used. The
glow was initiated by tuning the capacitor at maximum voltage to obtain an input current of
105 mA. The effect of power input to the oscillator on the glow discharge was determined
by varying the voltage. The change in glow is reflected in the current drawn by the oscil-
lator at any particular voltage and capacity setting. The following characteristics were
observed:
rn Increasing gas-flow rates reduces the glow intensity.
0 Increasing pressure reduces th'e glow intensity.
0 In general, glow intensity is reduced as the electrode distance is increased
Glows were maintained in the presence of a dielectric and a conductor filament to determine
their effect on glow characteristics. The presence of a strand of quartz, QFY-150 to 204
end, 1/0, with oil starch finish, d = 0.004 in. , did not affect the operating characteristics
of the glow. However, the presence of a 5-mil tungsten wire resulted in considerable
changes in the glow. The tungsten wire acted as a ground connection, and the ground lead
from the oscillator could be disconnected without affecting the glow. The glow was con-
centrated at and spread out along the wire. This system used much higher currents and,
as a result, considerable heat was generated. A mirror was deposited on the reaction tube
walls.
A series of experiments were performed using glass and quartz substrates to determine
approximate conditions necessary for the deposition of boron. A qualitative description of
the process was obtained from these observations. The data shown in Table I describes
the experimental conditions along with pertinent observations. The results have been
summarized in the following qualitative statements:
Wall deposition is favored by high hydrogen and low BCl3 flow rates.
rn Deposition on the filament substrate is more pronounced at higher flow rates of
BCl3.
0 Use of a water cooled reactor lowers wall deposition.
0 Small leaks of air result in the formation of the oxide rather than the metal (see
experiment 07-18A and Fig. 2).
0 Under certain conditions of power input and gas flow, a brown deposit is formed
that hydrolyzes on exposure to air and evolves a gas (see experiment 07-20A and
07-2lA).
0 Thick deposits of boron can be produced in reasonable times and conditions (see
Fig. 3).
Following these preliminary experiments, flowmeters were calibrated and the quartz mono- '
filament substrate was characterized in terms of physical and mechanical properties. A
study was then initiated on effects of the various process parameters.
Paramctric Study. A study of the effect of process parameters on the deposition of boron
using the static filament apparatus was conducted. The parameters studied include system
pressure, reactant gas flows and ratios, electrode separation and configurations, and field
strengths. The evaluation of certain parameters has been quantitative and waa baaed on
the weight of boron deposited or on the thickness of the deposit at the midpoint of the fila-
ment. Other parameters were evaluated by a qualitative deecription of the deposit.

v)
t-
at- InIo Io0 t-Q) loo lcua O 0) O ua I
0 0
3 2
oo o
m a

166
(a) Filament Emerging From Broken End (220~)
- W
-' - - ---
(b) Surface of Deposit at Broken End (220X)
Fig. 3 Boron on Quartz /
--
I

This study is only an approximation of the moving filament system. The movement of the
filament in a continuous system averages the effects of all the parameters so that the fila-
ment sample is prepared in a uniform manner. The edge effects noted on the filament
situated under the electrodes would not be evident in a continuous system.
Material efficiencies are based on the amount of boron produced from the total amount of
boron passed through the reactor. The electrical efficiency is based on the amount of
boron produced divided by the theoretical amount that could be produced by the total charge
passed wing the reduction equation B+3 - BO + 3 ~ The . current was measured by 811 rf
ammeter placed in the electrical circuit.
(1) Electrode Configuration
The experiments performed previously have used electrodes made from copper foil. The
electrodes are shown in Fig. 4. It was the purpose of one of the series of experiments
conducted to determine if other electrode configurations would be desirable. The configu-
rations shown in Figs. 4b, 4c, 4d, and 4e were used. The following qualitative observa-
tions were sufficient for determining the best design at this time:
The plate electrodes parallel to the gas flow shown in Fig. 4 produce a glow but
no boron deposition with gas flows and reactant concentrations known to deposit
boron when ring electrodes are used.
0 The large sheet electrodes shown in Fig. 4d encourage the deposition of boron on
the side walls between the electrodes.
0 Electrodes, using No. 16 copper wire, hooked up as in Fig. 4a produce a minimal
amount of sidewall deposition, and influence of the electrodes on the deposit under-
neath is also reduced.
0 Electrodes, using No. 16 copper wire, hooked up as in Figs. 4b and 4c produce
uneven glows and deposits. The glow will be intense between two of the electrodes
while the glow between the other electrodes is very faint. The conductivity of the
gas phases, electrode separation, and limitation of the current output of the oacil-
lator are probably important factors.
On the basis of the results observed, copper wire electrodes wound around the reactor tube
as in Fig. 4a have been used in the process study.
(2) Electrode Separation
The effect of electrode separation on the deposition of boron is show in Table II. The gm
flow, reactor pressure, and electrical parameters were the same in all cases. The data
show a rise in efficiency as the electrode separation is increased to 2.54 cm.
Electrode Deposit Deposit
distance length weight
(cm) Icm) (Ccgm/cm)
0.62 0.85 , 35.3
2.22 2.15 23.4
3.16 2.22 25.3
2.64(a) 2.15 46.6
Table II
EFFECT OF ELECTRODE SEPARATION
.
Total Materld
thickness efficiency emciency
1.8 0.34 I 0.76
1.67 0.69 I 1.62
1.72 0.66 I 1.41
2.22 1.16 I 2.66

la
lb H&i GASFLOW
GROUND
IC ‘-0 GAS FLOW
GROUND
GROUND
Id 0 GAS FLOW
le
GROUND
0 GAS FLOW
Fig. 4 Electrode Configuration
0
/
c

Mole Reaction Material
ratio time efficiency
BC13/H2 (min) (96 per pass)
0.22 7.80
0.19 5.0
0. 12 10.0
0.08 20.0
Electrical
efficiency
(% per pms)
0.046 1.2
0.044 1. 1
0. 06 1. 1
0. 14 1.2
As the mole ratio of BCl3 to hydrogen is increased from 0.22 to 0.66 (near 8toichiometric)
and by reducing the hydrogen gas fl~w, the appearance of the glow and the character of the
deposit are changed significantly. The hydrogen flow rate is an estimated value because
the calibration curve does not extend to these low flowmeter readings. However by cutting
the hydrogen flow rate in half, making the flow rate near stoichiometric for the reaction,
the glow is intensified around the filament substrate and will migrate along the subatrate
beyond the rf electrodes. The deposit character produced by this procedure ia shown in
Fig. 40. These pictures show the types of deposita that are formed at different points on
the substrate.
The deposition rate ig significantly affected by the flow ratio a8 shown in Table W.
Table IV
EFFECT OF FLOW RATIO ON DEPOSITION RATE
Mole flow ratio
(miWmin)
0.66 (est. )
-- _. . . -_ __ _. - -

1 6
170
Fig. 5 Filaments Produced at Near Stoichiometric Flow
Ratios (220 X)
1
4
I
\,

\
i
(4) Effect of System Pressure
The effect of system pressure on deposit characteristics is shown in Table V.
Pressure
(mm Hg)
5
20
30
(6) Interaction of Parameters
171
Table V
EFFECTS OF SYSTEM PRESSURE I
Type of deposit
Needles, trees, and dendrites.
Exploded from substrate if run was long
Thick, fairly uniform, some craters or knobs
Thick, small mounds, no craters
The gas phase being ionized is an integral part of the electrical circuit. Thus as gas com-
position, pressure, or electrode spacing is varied, the impedance of the gas phaee is ala0
changed. The necessary voltage and rf currents are thus changed. These changes are
reflected in the type of deposit obtained.
DISCUSSION
General. The results of these experiments show that boron can be deposited on a dielectric
substrate in a glow discharge. There is no limit on the coating thickness inherent in the
process. Complete control of the process parameters has not been achieved, but there is
no reason to believe that control cannot be achieved. The results obtained so far indicate
that gas purity, gas composition, and rf power input are among the most important factors.
Gas Purity. The deposit obtained can be considered to be in chemical equilibrium with the
gas composition. The introduction of certain impurities such as oxygen or water will result
in certain amounts of oxygen in the deposit. The solid solubility of oxygen in boron is prob-
ably not great, and its effect in the mechanical properties of the deposit is unknown. How-
ever, if a certain concentration of oxygen or water is present, the equilibrium solid phase
resulting from the processing will be the oxide and not the element. Experimental experi-
ence has shown that this concentration value is probably not very great, and a tight process
system must be used with the reduced pressure process.
Gas Composition. The effect of gas composition, i. e. , the ratio of hydrogen to boron tri-
chloride, will influence the type of reaction that occurs and the type of product formed. It
has been observed that the amount of (1) sidewall deposition; and (2) fine particles that fall
out of the gm downstream of the glow, increase with increasing hydrogen concentrations.
This indicates that redlrction in the gas phase increases with the higher hydrogen to boron
trichloride ratios. With increasing hydrogen ratios, the glow is more uniform over the
space and the formation of fine particles is encouraged.
Increasing the ratio of boron trichloride to hydrogen causes the glow to concentrate toward
I
'
the center mound the substrate. This concentration of the glow toward the centerdiscourages

172
the formation of small particles by the boron atoms and encourages the growth of the coat-
ing by deposition of the boron atoms directly on the substrate. As the flow ratio approaches
the stoichiometric value of 1.5 moles of H2 per mole BCl3, the glow concentrates very
close to the substrate and is observed to migrate along the filament beyond the electrode
rings. At the same time the rate of deposition increases about fourfold or more.
The resistance of the filament decreases as the boron coating is formed, which allows more
current to flow. This increases the temperature of the filament, which encourages more
rapid deposition of boron. Under these conditions, the reaction goes out of control and the
filament heats to a general red-orange glow with brighter spots and finally breaks.
Careful adjustment of the reactant gas concentration and the rf power input to the coil will
minimize the gas-phase reaction, sidewall deposition, and fine-particle formation.
The gas composition also obviously influences the nature of the material deposited. The
glow discharge of boron trichloride has been studied by several workers (Refs. 1, 5, and
8 to 16) in the field, and their results should indicate the deposit to be expected at high
ratios of boron trichloride to hydrogen. Holzmann and Morris (Ref. 1) analyzed the light
from glows of boron trichloride at 1 mm of Hg pressure. The glow discharges were
caused by a 2.4 to 2.5 x lo9 Hz (2,400 to 2,500 Mc). They observed bands due to boron
monochloride and lines due to atomic boron and atomic chlorine. Frazer and Holzmann
(Ref. 9) have used this technique to prepare laboratory quantities of B2Cl4. which is the
primary product under these conditions.
The compound B2Cl4 can be decomposed under a variety of conditions. Rosenberg (Ref. 8)
used 60-Hz 10-kV discharges in B2Cl4 and obtained BC13(g) and (BCl),, which hedescribed
ns a light brown or yellow film which is hydrolyzed by water to form B2O3, hydrochloric
acid, and hydrogen. From the rather incomplete data on this system and from other
authors, it appears that a (BC1)n phase can be obtained at compositions down to (BC10. g)n,
and a material which analyzes as BC10.6 has been reported. At the present time the (BC1)n
material is considered to be a polymer, and a series of higher unit structures other than
B2Cl4, (such as B4Cl4) can also be obtained.
In addition to these considerations, experiments have shown that, by operating at reduced
rf currents, a deposit possessing the reported properties of (BC1)n can be obtained with
gas compositions that would normally deposit elemental boron.
In summarizing the results and literature survey, it is possible by glow-discharge tech-
niqucs to obtain a variety of products from boron trichloride and hydrogen gas streams.
These products can range from a polymer-type (BCl)n, a liquid B2Cl4, or a solid BC10.6
to the boranes. In between these gas compositions that will give the foregoing products is
the gas-composition range that will give baron as a product. In addition to the gas com-
position, the amount of rf power used will also determine the product. It is fortunate that
the gas composition obtainable with our experimental apparatus is in the range for boron
deposition, otherwise selection of gas composition from present information would be
impossible.
Power Input. The power input to the glow is of great importance in determining the product
and character of the product. The rf voltage at any gas composition and electrode geometry
must be a value greater than the breakdown potential of the gases at that pressure. The rf
current is a direct measurement of the ionization of the gases to produce electrons and
positive ions. In electrochemical processing, these terms are or can be considered analogs
of the redox potentials and current density, respectively. At higher currents, a fastercoat-
ing rate will be attained, but changes in deposit characteristics will occur as a result of
such changes in plating rate. Perhaps part of the nonuniformity of the deposit can be
attributed to a nonuniform plating rate. The uniformity should be improved by moving the
filament substrate through the glow.
/
i

173
Pressure. Since pressure is a concentration function, it may be expected to influence the
type of deposit and the rate of deposition. At low pressures, needle-like deposits are
observed. As the pressure is increased, the type of deposit changes and a coating or continuous
layer is formed. Increasing the pressure further provides a smoother deposit.
However as the pressure is increased the voltage necessary to maintain a glow discharge
is increased.
I
Post-Deposition Treatment. One sample prepared at 30 mm Hg pressure, and at flowratios
and electrical conditions which normally provide a typical coating, was exposed to a hydro-
gen glow discharge before removal from the system. This was done to explore possible
methods of reducing the brown, glassy material observed near the end of filaments. The
treatment was done in 30 mm Hg H2 pressure and at low flow rates - 25 cc/min. The rf
current and dc voltage were 250 mA and 480 V respectively.
The glow was normal in every respect except that just outside the electrode on the down-
stream side a faint green fluorescence or glow was observed. The end areas of the sample
did show some evidence of reduction. The most interesting part of the sainple is shown in
Fig. 6. This is a longitudinal cross section of a fracture. Averythin (lessthan 0.187-mil)
layer is much darker than the bulk of the deposit. This has not been observed in any sam-
ples prepared during this process study.
FILAMENT EVALUATION
Density. The density of filaments produced by this process has been measured by two tech-
niques. A direct measurement was made by dropping a boron-coated pyrex rod in adensity
gradient tube. The average density was 2.24 gm/cc, which agrees satisfactorily with the
value estimated from the density of boron, 2.35 gm/cc, and pyrex, 2.24 gm/cc.
An alternate indirect measurement has been performed during the process study by com-
paring the observed thickness of samples with that calculated-from-weight measurements.
It was anticipated that the optically-measured value would be higher than the value calcu-
lated from weight measurements.which is an average value,since it is measured near the
center of the filament.
This’is observed in the data for samples 07-42 26A and,B. The total thickness measured
by optical methods are 2.2 and 2.4 mils, and calculated by weight the thicknesses are 1.45
and 1.72 mils. The results illustrate the nonuniformity of the static filament samples.
However, the data is sufficiently in agreement to indicate that the filaments produced
possess a density in accordance with the relative amounts of boron deposit and quartz
substrate.
X-Ray Identification. A typical x-ray analysis is shown in Table VI. The diffuse lines
observed in the x-ray pattern correspond to the alpha rhombohedral structure. This is
the low-temperature form of boron. However, the structure is not well developed as
indicated by the broad lines. The boron may have been in the process of changing from
an amorphous structure to a crystalline structure at the time the experiment was termi-
nated, or the crystalline boron was mixed with a matrix of amorphous boron.
Other samples of boron-coated 1.0-mil quartz filament show only two broad diffuse halos.
No lines indicating B2O3 were seen. No positive identification of the material was possi-
ble from the data.
Tensile Strengths. Several sample filaments of boron-coated 1-mil quartz substrate have
been tested for ultimate strengths. No attempts have been made to obtain an elastic modulus
since the samples prepared in the static apparatus are not of sufficient length. These
data are presented in Table VII. The coating cracked and in some cases completely exfoliated.
Thus, the values reported here basically represent breaking strengths of quartz
rather than composite filament of boron-coated quartz monofilaments.

Fig. 6 Longitudinal Cross Section of Deposit After
Exposure to H2 Glow Discharge
$ 4 -
I
! I-
@)
Fig. 7 (a) Longitudinal and (b) Cross Section of Boron-Coated
Quartz Monofilament (220 x)
-k
I

175
Table VI
X-RAY DATA OF DEPOSIT
Unknown intensity
(a) Alpha rhombohedral form.
Table WI
BREAKING WEIGHT OF SELECTED SAMPLES
Sample Sample condition Breaking Filament
Number prior to test weight (lb) diameter (mile)
07-40-24 Not observed 0.07 1.5
07-42-26 Not observed 0.07 1.5
07-52-36 Longitudinal cracks 0. 11 2.0
07-53-36 Coating missing near tab 0.033 2.0
Coating Evaluation. A typical sample cross section prepared during the process study is
shown in Fig. 7, demonstrating that acontinuous coating can be successfully deposited. In
the longitudinal cross section two mounds are fortuitously shown and it can be seen that
the growth extends to the substrate. This indicates that the cause of this different growth
originates at the substrate surface, possibly as a microcrack or particle of dust. The
problem of handling the filament and its contamination would be minimized with use of a
continuous system.
The bond between the coating and the substrate is essentially mechanical in nature. It
would be very desirable to improve this by effecting some sort of reaction between the two
materials to improve the overall properties of the filament. It would perhaps be undesir-
able to clean the organic or silicone finish from the substrate as has been done,
Microscopic examination of one of the filaments,has revealed a possible explanation of the
reduced strength of boron-coated quartz filaments. The pictures in Fig. 8 are typical of
what is occasionally seen on samples. The chip and hole in the top and middle picture
were not measured, but the rather deep cut made indicates that the quartz monofilament
possesses weak portions that may separate before the bond between the boron and the
quartz ruptures. These weak volumes contribute to the spread of the strength values
observed earlier since they are probably randomly distributed. The hole and ball shown
in the bottom figure have been measured during the microscopic examination. The ball
W- &at 1.0 mil in diameter, the hole WBB observed to 0.8 mil, while the filament cram
eection averaged 1.8 mils. The coating thickness can be only 0.4 mil, so that the hole
must have been formed by the loss of 0.4 mil of the quartz substrate.

4
176
Fig. 8 Boron-Coated Quartz Monofilament (x 340)
Top - Hollow in Boron-Coated Quartz Monofilament
Middle - Material From Hole
Bottom - Ball and Hollow in Thick Coating on Quartz
Monofilament
1
i

177
Another source of weakness is the cracking and peeling. Many of these cracks and breaks
are the result of handling during removal from the static apparatus. Use of a continuous
filament moving through the apparatus and improvement in the adhesion of the Coating to
the substrate would possibly eliminate this problem.
CONCLUSIONS
This paper basically represents a feasibility study of the deposition of d o n from a glow
discharge system produced in a boron trichloride-hydrogen medium. Feasibility of the
basic concept has been conclusively demonstrated for dielectric or insulating substrates.
It is probable that conditions can be adjusted to provide for deposition on metallic or conductive
substrates as well. The following secondary conclusions can also be drawn from
this work:
0
0
'0
0
e
0
0
0
0
Glow discharges can be initiated a d sustained in hydrogen-boron trichloride mix-
tures at pressures of 50-100 mm.
There is no apparent limit on coating thickness which can be deposited.
The deposited boron is amorphous with a partial alpha rhombohedral crystalline
character.
Deposition rate is of the order of 5 x 10-5 gm/sec.
Filament resistance increases with thickness; temperature in turn increases
resulting in a higher deposition rate; this also results in increased bodning to the
substrate by chemical or diffusive action.
The process is quite sensitive to small amounts of impurities in the gas stream,
particularly oxygen and moisture.
The ratio of hydrogen to boron trichloride has a major effect on nature of the
deposit; both sidewall deposition and downstream particle fallout increase with
hydrogen concentration. However, the glow is more uniform at high hydrogen
values, so these factors must be balanced for optimum performance.
An increase in the boron trichloride to hydrogen ratio tends to concentrate the
glow around the substrate, promoting efficiency and minimizing extraneous
deposition.
High hydrogen to boron trichloride ratios tend to produce boranes; operation at
reduced rf currents tends to yield (BCl), polymers.
Current investigations are concerned with further development of the glow discharge deposi-
tion procedure to operate on a continuous basis utilizing moving filament techniques. Indica-
tions are that the technique may be used to prepare a high-quality boron filament at
reasonable processing rates.
ACKNOWLEDGMENT
This work was sponsored by the U.S. Air Force under Contract AF 33(615)-2130. Acknowl-
edgement is given to J. G. Bjeletich, W. C. Coons, J. Robinson, B. A. Traina, R. N.
Varney, and R. D. Wales for assistance on photographic and experimental portions of the
program. All are members of the Lockheed Palo Alto Research Laboratory. Further
acknowledgment is made to R. M. Neff of the Air Force Materials Laboratory for con-
tinuing consultation on progress of the program.
1.
2.
3.
4.
REFERENCES
R. T. Holzmann and W. F. Morris, J. Chem. Phys. , Vol. 29, 1958, p. 677
W. V. Kotlensky and R. Schaeffer, J. Am. Chem. Soc., Vol. 80, 1958, p. 4517
R. M. Rosenberg. Univ. Microfilms (Ann Arbor, Mich. ) L. C. Card No. MIL-59-2911.
160 pp: Dissertation Abst., Vol. 20, 1959, p. 526
L. Ya. Markovskii, V. I. L'viva, and Yu D. Kondrashev, "Obtaining Elementary
Boron in a Glow Discharge." Bor. Trudy Konf. Khim, Bora; Ego Soedineii, pp. 36-45
(1958); cf. FTD-MT-64-427, Foregin Technol. Division, AF System Command, 1985

5. F. Weintraub, Trans. Am. Electrochem. Soc., Vol. 21, 1909, pp. 165-184 ,
6. W. Kroll, 2. Anorg. Allgem. Chem., Vol. 101, 1918, p. 1
7. "Report of the International Symposium on Electrical Discharges in Gases," Delft,
Netherlands, April 1955; The Hague, 1955
8. R. M. Rosenberg, PhD Thesis, Pennsylvania State University, 1959
9. J. W. Frazer and R. T. Holzmann, J. Am. Chem. Soc., Vol. 80, 1998, pp. 2907-2908
10. T. Wartik, R. Moore, and H. I. Schlesinger, J. Am. Chem. Soc., Vol. 71, 1849,
p. 3265
11. G. Urry-, T. Wartfk, and H. I. Schlesinger, J. Am. Chem. SOC., Vol. 76, 1954,
p. 4223
12. A. Stock, A. Brandt, and H. Fischer, Ber. Deut. Chem. Ges., Vol. 58B, 1925, p. 653
13. H. I. Schlesinger and A. 3. Bing, J. Am. Chem. Soc., Vol. 53, 1931, p. 4321
14. A. Stock, H. Martin, and W. Sutterlin, Ber. Deut. Chem.. Gee., Vol. 67, 1934, p. 396
15. F. F. Apple, PhD Thesis, Pennsylvania State University, 1955
16. H. I. Schlesinger, H. C. Brown, BI Abraham, N. Davidson, A. E. Finholt, R. A. Lad,
J. Knight, and A. M. Schwartz, J. Am. Chem. Soc., Vol. 75, 1953, p. 191
L

179
PLATING IN A CORONA DISCHARGE
R. D. Wales
Materials Sciences Laboratory
Lockheed Palo Alto Research Laboratory
Palo Alto, California
ABSTRACT
It has been demonstrated that boron can be deposited on a suitable
substrate in a corona discharge. Application of high-voltage electri-
cal energy across a gaseous hydrogen-boron halide mixture forms
ionized/activated states, resulting in the deposition of metallic boron.
The deposition process is electrochemical in nature, the boron being
deposited cathodically.
INTRODUCTION
Literature on electric discharges is quite extensive. However, most of this iiterature concerns
the nature and properties of discharges taking place in stable as systems. Dctailed
discussions of electrical discharges are presented in several books(f* 2* 39 4). Relatively
little information is available concerning chemical reactions initiated and sustained by
clectrical discharges. Arc reactions are basically thermally initiated reactions deriving
their thermal energy from the arc and thus are not applicable in the present sense. Glow
discharges have been utilized for the polymerization of organic materials and the deposition
of oxide films. The formation of pure boron powded5) and the decomposition of diboratd)
and boron tri~hloride(~) have been studied in a glow discharge. No references have becn
located concerning chemical reactions initiated and sustained by a corona discharge.
Those references(*$ 99 lo) which refer to the utilization of a corona discharge to catalyze,
or cause, a chemical reaction do not in fact utilize a corona discharge. The discharge
utilized in these references has an odd resemblance to a glow discharge, but is actually a
suppressed spark. Thus, the discharge goes through the phenomena of spark buildup without,
however, generating a spark. Direct current cannot be used in these systems since
the buildup of charge at the insulated electrodes quenches the discharge, and a reverse
cycle is necessary to remove this charge and reinitiate the discharge.
A corona discharge is, essentially, intermediate between a glow and an arc discharge
being, however, a low current phenomenon. Glow discharges are generally generated at
low pressures while an arc is generally a high-pressure phenomenon. The gas is ionized
more or less uniformly throughout the system in a glow discharge. An arc is obtained
when a narrow path of ionized particles is generated between two electrodes, resulting in a
low resistance path which further aggravates the situation until extremely energetic particles
exist in the narrow path. A corona discharge is not as pressure-dependent and is obtained
at a small electrode which is opposed by a much larger electrode. There is a very large

160
change in field through the corona, but very little change over the remaining distance to the
opposing electrode. Most studies of corona discharges have been in inert gas systems, and
have been generated at a point electrode.
This study has been directed toivard the tletcrmination of the feasibility for depositing :1
coating, c.g., boron, on a suitablc substrate in :I corona discharge. A further object has
becn to determine some of the critical pnramctcrs and a possible indication of the mechanism
and characteristics of this technique for coating the substrate material. ;
EXPERIMENTAL WORK
A schematic diagram of the system used in this study is indicated in Fig. 1. The system
pressure was controlled with a Cartesian Manostat GA (Manostat Corporation) and a Welch
mechanical pump Model 1402 (W. M. Welch Manufacturing Company). The discharge was
initiated and sustained with a 7,000/12,000 V Jefferson luminous tube outdoor-type trans-
former (Jefferson Electric Company) connected to GO-cps 110-V power source through a
Variac variable transformer.
To dcterrnine the effect of alternating and direct current, an RCA CR 212 half-wave recti-
fier \vas added to the circuit. At the same time a resistance bridge, a current measuring
rcsistor, and an ammeter were added to the circuit. A schematic of the circuit is presented
in Fig. 2. The resistance bridge consisted of six 2-W 2-megohm resistors (R2) and one
2-\V 1,000-ohm resistor .(R1), all in series; the total voltage being 11,775 times the voltage
across the 1,000-ohm resistor. The current measuring resistor ,was a 2-W 10.075-ohm
resistor. The voltage measurements were made with a Hewlett-Packard Model 130B
oscilloscope, and alternating current measurements were made with a Simpson Model 378
milliammeter.
The reactants utilized in this study were hydrogen, helium, or argon, and boron tribromide.
The hydrogen was passed through a ffDe-oxo'7 unit (Engelhard), through a drying column of
magnesium perchlorate, and then through a flowmeter. After the flowmeter, the hydrogen
was split into two streams, one of which was bubbled through the boron tribromide, and
then recombined into one stream before entering the reaction chamber or cell.
The temperature of the bubbler, and of the cell, was controlled by wrapping each (and the
associated gas lines) with heating tape and controlling the power input with a Variac variable
transformer.
The system was maintained at or below atmospheric pressure for this study, with most of
the data being obtained at 2 to 3 in. of mercury below atmospheric pressure.
Tungsten wire was chosen as the substrate material, This wire was cleaned by passing it
successively through a train consisting of concentrated nitric acid, a distilled water rinse,
and an acetone rinse.
Using the hydrogen-boron tribromide reactant system and 60-cps alternating current to
develop n corona, approximately a dozen cell designs were tried. The cell design which 1
appears to be the most satisfactory is pictured in Fig. 3. This cell consists of a 2-in.long
pyrex glass tube with a neoprene O-ring inside each end and wrapped with eight turns
of 3-mil tungsten wire along the tube length such that the wire is parallel and concentric
to the axis of the tube. The wire is 0.9 cm from the axis. This assembly is then inserted
in a 5-in.-long pyrex glass tube 1.5-in. in diameter and supported such that the tubes are
concentric. A tungsten wire lead is then extended out of the tube, and neoprene rubber
stoppers inserted in each end. Each stopper has a capillary tube through its center such
that there are approximately 3 in. between the capillaries, and the filament enters and
leaves the cell through these capillaries. The reactants also enter (and leave) the cell
i!
through the stoppers.
I
4
1
,
I

\
3
,
,\
110 v
60 CPS
MANOMETER .
Fig. 1 Schematic of System Utilized for Plating in a Corona Discharge
7 300 TO 15 000 V
TRANSFORMER
VARIAC
MILLIAMMETER
Fig. 2 Schematic of Power Input and Measuring Circuit Used With the Corona
I' Discharge Plating Apparatus
i
I
I

Deposition Utilizing Altcrnating Current
1s2
PRIAIXRT \'XRI;\DI,E COXSIDERATIONS
Irydrogcn and boron trihronude vapors iverc used to obt:iin a co:iting of boron on 1-niil
tungsten irirc in a corona dcvelopctl ivith (X)-c.ps alternating currcnt. It \vw determined th;it,
at about 70 and 350 mA/cni2, no coating is ol,t:lined irhen the cell (and reactants) are at room
tcniperature, \rhilc a co:iting of boron is obtnincd if the temperature is raiscd to approximately
100" C. At highcr currcnt densities, the coating is concentrated in nodules. If enough power
is supplied to heat the filament to a dull red glow, :I good coating is obtained in the hot zone,
At lower current densities, a more uniform coating is obtained, depending upon the reactant
ratios and flow rates. Thus, hydrogen was bubbled through boron tribromide at a rate of
approximately 400 ml/min, and the system \viis maintained at approximately atmospheric
pressure. When the boron tribromide \vas maintained at about room temperuture and at a
total hydrogen flow rate of about, 1.5 liter/min, thc corona was discontinuous along the wire
and boron deposited in the area of the coronas to give discrete coated and uncoated areas.
With a hydrogen flow rate of about 400 ml/min and a plating time of about 10 min, a much
smoother coating was obtained. There was some tendency for the corona to break into
discontinuous sections, which resulted in a nonuniform buildup of boron.
When the boron tribromide was heated to about 30" C and the hydrogen flow rate was main-
tained at 400 ml/min, the corona was somewhat more continuous and the coating nearly
uniform. However, the coating was not nearly as smooth as in the previous example.
There was also a tendency for the formation of corona points on the opposing electrode
system with a resulting deposition of boron under the corona.
Deposition Utilizing Direct Current
Using hydrogen and boron tribromide as reactants to obtain a deposit of boron on 1-mil
tungsten wire in a corona developed with half-wave rectified 60-cps alternating current,
no deposit was obtained when the 1-mil wire was the anode. However, a coating was
obtnincd when the 1-mil wire was the cathode. Good, continuous corona and boron deposits
were obtained when the corona was initiated in a large excess of hydrogen (the hydrogen
flow was then decreased to the desired amount after about 5 min).
The coating is quite uniform and has an orange peel appearance at 2,OOOX magnification.
Some of the morphology of the substrate is apparent at high magnification (e.g., die marks).
The voltage requirements for a particular current increased as the hydrogen flow decreased,
increasing from more than 1,000 V peak to about 7,000 V peak as the excess hydrogen was
decreased in these runs.
Attempts to utilize pure de power were unsuccessful. The power supplies available were
designed to deliver several hundred milliamperes and had relatively coarse controls. Thus,
as the corona was initiated, the power supplies did not allow sufficient control, resulting in
"runaway" or burning in two of the substrate wire.
Deposition Utilizing Argon and Helium
Using helium instead of hydrogen, a coating was obtained, but arcing was much easier and
thus lower current (and lower flow rates) were necessary. Peak voltages of 7,000 to 8,000 v
were used.
Using argon instead of hydrogen required higher voltage (with lower currents) and resulted in
the formation of corona points under which boron was deposited to give whiskers. Peak
voltages of about 9,000 V were used. Whiskers greater than 6 mils long and about 0.5 mil
in diameter were obtained before discontinuing these experiments.
I
.1
I!

I
i
, \
I
i
& a -
P
Fig. 3 Reaction Cell Used for Plating in a Corona Discharge
Fig. 4 Boron Deposit on 1-mil Tungsten Wire, Run 28-1,
Tungsten Substrate Material at Top (340x)
Fig. 5 Boron Deposit on 0.5-mil Tungsten Wire,
RLUI 35-1 (450X)

184
Bromine was a product when helium or argon was used while hydrogen caused the formation
of hydrogen bromide.
Discussion
jl) Cell Design. The reaction cell should be designed so that the concentration of reactants
and products is essentially constant throughout the cell to give a more uniform corona and
deposit. The electrode geometry and separation are to some extent depended upon the power
requirements, which are in turn dependent upon the resistance heating effects in the filament
and upon the arcing probability. Thus, in practice, the current in the filament should not be
great enough to heat the filament excessively. With this current limitation, the electrode
separation is dependent upon the arcing probability and should be great enough to'level out
any small variations in electrode separation which could cause the current to concentrate in
a small region. Furthermore, to obtain a unfirom corona around the filament, the opposing
electrode should be concentric around the filament. If the power input is not pure direct
current, a "motoring" effect results if the opposing electrode is a wire coiled around the
filament. That is, a vibration or circular oscillation will be induced on the filament. There-
fore, the opposing electrode has been made up of a group of connected electrodes arranged
concentrically around the filament and parallel to it in order to minimize the "motoring"
effect. The opposing electrode could also have been a metal cylinder or screen, but in order
both to observe the filament and be able to operate at reasonable electrode separations
(reasonable voltages), interconnected parallel wire electrodes were selected.
12) Mechanism of Deposition. The results obtained indicate that the mechanism of boron
deposition in a corona discharge is cathodic. In this type system, electrons are generated
and repelled from the cathode while positive ions are drawn to it. At the anode, positive ions
are generated or electrons are drawn to it. Thus, the boron radical is positively charged and
drawn to the cathode where it is deposited as elemental boron.
Helium and argon were used to determine if the boron tribromide was ionized directly, or if
its ionization was a secondary reaction. Table I indicates some of the ionization levels for
these gases. Helium in a corona discharge has only one or two potential (or ionization) levels,
and these are quite high. However, helium is ionized at lower applied voltages than most
gases because the electrons generated at the cathode have only elastic collisions with helium
atoms until they acquire enough kinetic energy from the field to ionize the helium. Argon has
a potential (or ionization) level which is relatively low (about half that of helium) but higher
than the ionization levels for hydrogen. In helium the system was very susceptible to arcing,
while in argon higher voltages were required with some susceptibility to arcing. In hydrogen
arcing was not such a problem because of the many types of inelastic collisions possible and
the consequent tendency to decrease the electron energy and concentration. Boron was
deposited in all three systems. However, in hydrogen, hydrogen bromide was obtained while
bromine was obtained in both argon and helium. Deposition was slightly easier and there
was a slightly increased rate of deposition of boron with an increased tendency to arcing in
helium and argon. The difference, however, was not sufficient to be due to direct reaction
with the free electron, and indicates that the boron tribromide is ionized by collision with
ionized particles rather than electrons or the field. Thus, the mechanism includes: (a) the
ionization (or activation) of the hydrogen (helium or argon), @) the collision or exchange of
energy of the ionized particle with boron tribromide to result in (c) the formation of a
positively charged boron radical and hydrogen bromide (or bromine), (d) the transport of
the charged boron radical to the cathode in the field, and (e) the electrochemical deposition
of boron on the cathode.
It is apparently necessary to add a small amount of kinetic (thermal) energy to aid the
deposition in hydrogen. However, this reactant system offers certain advantages Over
helium or argon systems. That is, with hydrogen, there are apparently more ionized or
activated particles with fewer electrons and consequently less likelihood of creating
conditions conducive to arcing.

\
\'
C
0
.d *
d
.-
C
U
d
0)
c(
1)
L-
. . . .
I n w I n c u
O d d 0
N d d d
e 8 m

125 I
I The C'orol;:~. In tiic Iiigii-iicld rcfiioii ni':~~' the' sul,..;trnte wire. positive ions can nttain
high encrgics :ind ttic sccond:ir!- clc~ctron cmission process is quite clficient. 'rhese
secont1:iry electrons \\-ill start electron avalanctics. \[.hick \vi11 iti' turn produce rna~iy:positive
I
I
I
ions, \vhich protlucc niorc :cv:il:inchcs, ctc. ;\t lo\\ ~)rcssur~'s, tlilf~ision ol thc electron
avatanctics sprcati..; thc glo\v anti tiistril)utcs !lie iiositive sixice cti;irge so that ihc discharge
is not yucnchcti. 1)ifiusion docs not talic p1:icc at higher pressure, and ;I localized dense
!
sixicc charge estinguishcs 'thc corona. 'l'he extinction lnsts until the positivc space ch;!rge
tiiffuscs to ttic clcctrode and the last ions rcinitiiite ttic ~ischargc, resulting in ;t periodic
coron:i \vith a frequency dependent upon the ficltl ~uid the velocity of thc positiveions.
1
\\'itti the ncgativc wirc (cathode), the ionization increases first with distance from the wire,
rcaches ;I ~iculr, then drops sharply. ?'hat is, the electrons move out'from thc wire and
produce relatively stationary positive ions I)ct\vecn thcniselves and the wire, thus weakening ,
the field at greater distances. The ninsimum ioniz:ition occurs ;it several ionizing; free paths .
from the \vim, antl there is ;I dark space Iict\veen thc wire and the luminous glow. Furtherniorc,
as mentioned, at low pressures the lxtcrnl diffusion of electron avalances spreads the
coron;~ over the wirc surfncc. Uecausc the vnluc of the coefficient of electron epission is
iniport:int. antl because the glow will not be uniform if the coefficient of electron emission
varics over the surface of the wire, the glow may settle in p?tches of greater or lesser , '.
luminosity and at high pressures will appear :IS a single small area of.corona. The corona
niny cshibit a marked flickering near threshold because the positive ion bombardment :may
changc the effective coefficient of electron emission by denuding the surface of its gas film..
The discharge thus decreases or ceases in that rcgion, resuming again when the surface Ins
recovered. At low pressures and voltages well above the threshold value, the corona is fairly,
steady.
In this work it was necessary to clean the substrate wire so that there were no variations in
thc,coefficient of electron emission along the wire, and thus.a localization of the corona into'
points. During deposition the corona was distributed over the substrate wire in a uniform
manner. However, the discharge was periodic as indicated in the above discussion. That is,
the shcnth (of light) dong the wire was not of a uniform brightness, but vnsisted of brighter
nncl darker areas which seemed to move along the wire in a somewhat random manner,ibut
which appeared periodic in nature at a given point on the wire.
I.:s pe r i men t a1 \V or k
SECONDARY VARIABLE CONSIDERATIONS
The deposition system was that previously described. The tungsten wire was cleaned as
before, thcn placed in the cell. The corona was then initiated in excess hydrogen and mintained
for about 5 min, or until the corona was continuous or nearly continuous along the
substrate electrode, before the conditions were changed to give boron deposition. This
initial "cleaningt1 process was effected with hydrogen flowing at about 16Occ/min through
the boron tribromide at about 26" C and with hydrogen flowing through the bypass line at
about 270cc/min. The electrical input during this time was about 17 mA peak and about .
4,700 V peak for 1-mil tungsten wire and about 5,600 V peak and 4.5 and 11.0 mA peak
for 0.5-mil tungsten wire.
Results
11) Deposition on 1-mil Tungsten Wire. Some of the results obtained are presented in
TaMc 2. The corona was maintained continuously in all run.s. A picture of the fihment
olitained in Run 28-1 is shown in Fig. 4. In Run 28-1 the current fell from an approximate
initial value of 12.3 mA peak to about 3.. 5 mA peak in about 9 min, after which it was
nearly constant. The peak voltage changed from 6,900 to 13,190 V. The filamedC obtained
in this run is apparently the desired type, although not of optimum thickness.
,
. '
'
1
4
.J
6.
. , i
I

I
i
I .
Q)
I,
s
w
0
v)
Y
0
a
w
W
a
!
rn
cu
I
0
W
rl
187
c) In I SI SI
0 0 * 00 00 00
*In .ma

188
obtained in the experiments presented, and in other similar experiments, indicate
the voltage requirements increase as the boron tribromide to hydrogen ratio
increases.
The voltage requirements increase as the system pressure increases.
The substrate is etched with little or no boron deposited as the boron tri-
bromide to hydrogen ratio decreases.
The corona tends to break up into points at high flow rates, resulting in nonuniform
boron deposits.
Good coatings were obtained at high boron tribromide to hydrogen ratios and relatively low
currents.
12) Deposition on 0.5-mil Tungsten Wire. Some of the results obtained are given in Table 3.
The corona was maintained continuously in Runs 30-1 and 30-2. The power input was dis-
continued in all remaining runs until the boron tribromide and the cell has attained the desired
temperature. The currents and voltages indicated for all runs below 35 are approximate
since no adjustments were made once the run had begun.
The effects of operating conditions are similar to those observed for deposition on 1.0-mil
tungsten wire. However, the decreased heating effects of the 0.5-mil wire necessitate heating
the cell to about 200°C to prevent condensation of the boron tribromide.
In Run 31-1, an apparently desirable coating was obtained at the ends of the reaction zone
while the central portion of the filament was quite noddar. In Run 31-1 the peak current was
6.2 mA, falling to about 1.8 mA at the end of the run. During this time the peak voltage
changed from 12,480 V to about 15,780 V.
Runs 35-1 and 35-2 yielded filaments having uniform coatings of boron. The coating, as
shown in Fig. 5, had ridges along its surface resembling die marks on wire. Figure 6 shows
a cross section of filament from Run 35-1 in normal and in polarized light. The higher magni-
fications indicate that the markings on the surface are reflections of the markings on the sub-
strate in almost a one-to-one ratio. Furthermore, there is no indication of boride formation.
However, polarized light indicates some anisotropic properties similar to pyrolytic graphite.
The material seems softer and much darker than that grown by gas plating techniques. Although
there are some unidentified lines, x-ray data indicate that the coating is amorphous boron with
no borides in the filament.
Cursory examination of the surface for a few of the substrates has indicated surface finishes
varying from the ridged (die marks) through smooth to an orange-peel-type finish, while the
filaments obtained have had surface finishes ranging from ridged through smooth to nodular.
Therefore, the substrate surface is a critical factor in determining the appearance and sur-
face of the final filament.
13) Effect of Morphology of the Substrate. The 0.5-mil tungsten substrate wire has been
examined after various stages in the process. Some of the results are presented in Table 4.
It has been determined that the liquid cleaning train cleans and slightly etches the wire,
and the high hydrogen corona etches the wire as indicated in Table 4. At low peak currents
the die marks were etched away, but the surface has a very rough orange peel finish. Using
both high and low currents consecutively, a much smoother orange peel finish is obtained.
Boron deposited on wire etched to an orange peel finish gives a coating similar to that
pictured in Fig. 7. The morphology of boron deposited on variously etched wire verifies the
hypothesis previously presented in that the coating morphology and growth structure are
directly dependent upon the morphology of the substrate.
1

I
(b) Polarized Light
Fig. 6 Cross Section of Filament Obtained by Plating in a Corona
Discharge, cf. Fig. 5 (3,000~)
I

0
5 .d
3
v)
*
.d
3
0
u
a 6
.d E *g
e-
3%
SP)
0 a-
>
0 00
e ' (0
I
*
W
Q) d
d
rl
rl
t-
CD
I
In
E-
In 0
m w
In
r(
d N
m
In
O E-
c) . h l
0 d
N
03 m
9 . (D
0
N
N
N
0 0 m 0 0
dl rl d m rl
0
u)
0
v)
0
v)
0
In
0
In
I0 (0 E- In
c-
01 c) ea N
N
I
1

i
m
v)
Y
C
i!
u
o)
23
a a-
+
191
m m 0 v)
* * * P)
v)
0
N
~
N
m
0
@a
*
0
01
m
W
m
N
*
v)
v)
@a
@a
@a
v)
OD 0 *
@a
@a
0 0 9
hl 4 OD
0
m
r(
I
m
0
m
0
m
0
N
0 In

192
(a) 604x
Fig. 7 Boron Deposit on 0.5-mil Tungsten Wire, Run 52-1,
cf. Table 4

i,
d
C
.d
c c
W Q ) 0 0
L1Y
0 000 00 00 ov) 0
t- t- P-t- t-t- t-e t-m
8 8 .z .z N N NN NN NN N

N
LI Y
0)
ucd +
194
mol 0
rl
7 s I I
4
(D 1
m
(D
i
I?
1

i
e 4 0 wcr
-4
4
al
0
0 0
c)
cr 13:- I
m
‘ E
100
Q)
uao
Q)v)
I
0
P-
cu
0
(0
l-l
195
w e N N m (D
rl rl
0
a
0
C
. w
x
E
._
E:
0
c Y
a,
4
k
al
5
o e o
Ncua
CJ
cu l-l d
I
I
d
m
P-
P-
l?
CJ 00
cd dc;
mu3
4P)
Y
cd

(-1) \'ariation of 1'l;iting Conditions. Comliining thc iniorm:ition obtained in the previous tests,
several runs irere macle using.-vnri,:itions of the- pl;itinfi conditions.
ohtainctl are prcscnte!l in Table 5 and in Fig. 5.
Some of thc, results
The current-tinic :intl.voltage-time chnr:ictkkistics for t\ro runs :ire indicated in Fig. 8.'
TIIC fil:imcnt oI1t:iincd in Ilun 73- 1 (T:il)Ie 5) \vas cvenly and uniformly coated similar to
thc filamcnt picturcd.in Fig. 5. Except for oiic defect, the filnmcnt obtained from Run
74-1 (Tablc 5) \v:is evenly and uniformly coated similar to the filament pictured in Fig. 7.
As indicated in Fig. 8, the defect probably occurred after an elapsed time of ?bout 21 min
at a cell voltage of about 23,600 V.
Thc results obtained.indicate that the best process conditions and plating procedure for
boron deposition on a stationary 0.5-mil'tungsten !\ire, are as indicated in Fig. 8.
Discussion
Thc morphology of the deposit obtainecl reflects, in general, the morphology of the sub-
strate. 'I'hc coating is essentially amorphous in nature although there is indication of
anisotropy.
One set of operating conditions giving ;I .satisfactory coating of boron on tungsten is presented
graphically in 13g. 8. In general, good coatings have been obtained at high boron tribromideto-hydrogen
ratios and relatively low currents. Furthermore, if the flowrate is too great,
the corona tends to break up into points, resulting in nonuniform deposits.
Other effects of varying operation conditions which suggest limitations for deposition, and
which also tend to verify the mechanism for deposition .. .. presented previously, include:
. .
(a) The voltage requirements increase as the boron tribromide to hydrogen ratio
increases.. ._
(b) The voltage requirements increase as the system pressure increases.
(c) The substrate is etched with.little or no boron deposited as the boron tribromide-
to-hydrogen ,ratio decreases. '..
CONTINUOUS. BORON DEPOSITION ON A MOVING FILAMENT
,.
Experimental Work
.. . . ._
A system was designed, assembled, and tested for. continuous boron deposition on a moving
filament. Figure 9 is a schematic of the apparatus. The design resembles that for the
batch plating process (Fig. 1);. ,The filament is pulled through the system with a Graham
constant speed motor. The spool is mounted ,on a shaft and suspended in;such a manner as
to offer minimum friction.. The seal and,electrical contact between cells is a unique arrangement
such that the filament moves through the seal in a vertical direction without breaking
.. . . ., I'
the seal.
: I .I. . j ..
The power for the ceils has been obtained'in a manner similar to that used for the batch plating
process (big. 2). There has been a, considerable problem with 60-cycle pickup in the measurement
system, Shielding;-ahd grounding, at various critical positions were necessary to elimin-
. I
ate this problem. 2. .
IIydrogen-boron trichloride was 'chosen ab the'reactant 'system'. The reactant mixture has
bccn obtained by bubbling hydrogen through boron trichloride maintained at about -2O'C.
Thc reaction cells havemot been heate$ externally, and the vapor flow was split between all
three cells. .,. ... .
The etching cell was n& Zil as'such;'arid Was eventually converted to use for cleaning )
the substrate filament by the hot-wire technique. Thus,, the chemical cleaning train was
eliminated. Hydrogen only was passed through the cleaning cell and the substrate heated to I
, . . _'. I, .. "3:. :_
~ ,,. _. , . ..
t
i
6

\
E
‘ 1
TIME FOR RUN
13-1 14-1
- 1
0 0
a 1
b 2
c 3
d 4
e 5
H2 FLOW
BUBBLER
(M L/MIN)
160
160
95
50
50
e
H2 FLOW BBr3
BYPASS TEMPERATURE
LML/MW (* C)
270 -22
270 Htrs. on
. 95 -35
0 -50
0 -72
Shut down
I I 1 I I I 1 I
4 8 12 16 20 24 28 32
PROCESSING TIME (MIN)
28
24
S
2o 3
B
16 0”
Fig. 8 Electrical Characteristics During Deposition of Boron on
0.5-mil Tungsten Wire in a Corona Discharge
12
E
4
w’
0
4
s
cl
w
u

H2 --1
DE-OXO DRYING
TOWER
TAKE-UP REEL
CHEMICAL
CLEANING TRAIX
COLD TRAPS
Fig. 9 Schematic of Continuous System for Plating in a Corona Discharge
Fig. 10 Boron Deposit on 0.15-mil Tungsten Wire,
Continuous System (600~)
i

199
a red heat in this cell using dc power. Tests performed in this system have utilized both
0.5-mil and 0.15-mil tungsten substrate.
Results
11) AC Pickup. The use of shielded leads and wire cages around the transformers has,
with the use of 1,000 ohm current measuring resistors, virtually eliminatyd the ac pickup.
j2) Dcposition on 0.5-mil Tungsten IVire. Satisfactory results have been obtained at a
hydrogen flow rate through the bubbler of about 100 cc/min and additional hydrogen at a
flow rate of about 350 cc/min. Hydrogen is flowing through the cleaning cell at about 140 cc/
min and the wire is cleaned by resistance heating at a dc voltage of about 70 V and a DC
current of about 320 milliamperes. Some of the results obtained are indicated in Table 6.
Using only one cell, at a filament speed of about 3.6 in. /min a coating about 0.25-mil
thick was obtained; at a filament speed of about 14.4 in./min, a coating ab-out 0.05-mil
thick was obtained.
Using two cells, there was some problem in maintaining the corona in the second cell. At
a filament speed of about 3,6 in. /min a coating about 0.7-mil thick was obtained in the
areas where a good corona was maintained. At a filament speed of about 14.4 in./min and
a current input to the second cell of about half that to the first cell, a coating about 0.07-mil
thick was obtained.
13) Deposition on 0.15-mil Tungsten Wire. The 0.15-mil tungsten substrate wire was cleaned
'
by the hot-wire technique at hydrogen flow rate of about 140 cc/min and at a dc power input
of about 208 V and 90 mA. While the input vapors were split between the three plating cells,
only one cell was utilized for plating studies. Some of the results are indicated in Table 7.
The best results were obtained at a filament speed of about 5 in./min with a hydrogen flow
rate through the bubbler of about 70 cc/min, and additional hydrogen at a flow rate of about
350 cc/min. To maintain a uniform corona, it was necessary to switch the power on and
off relatively slowly during the run. Under these conditions a coating thickness of about
0.17 mil was obtained.
There is apparently some difference in either the morphology of the 0.15-mil substrate
as compared to the 0.5-mil material, or less effect by this substrate upon the morphology
of the deposit. Although the morphology of the material deposited on 0.5-mil tungsten
appeared similar to that indicated in Fig. 5, the morphology of the deposit on the 0.15-mil
tungsten appeared more like that indicated in Fig. 4. Figure 10 indicates the morphology
usually obtained 0.15-mil tungsten substrate. The total diameter of this sample is 0.4 mil.
Discussion
An apparatus has been presented, and tests performed, indicating the feasibility of depositing
boron on a continuously moving substrate in a corona discharge.
Although no tests were made for verification, the results obtained with the 0.15-mil sub-
strate indicate that the substrate diameter also effects the operating conditions, probably
though buildup of the positive ion sheath to such an extent that the corona cannot recover
after being quenched. This effect is possibly related to the similar effect of high flowrates
on larger substrates noted in the previous section. The effects might be explained by postu-
lating a higher concentration of boron radicals around the smaller substrate simply through
volume considerations, and around the larger substrate through an increased concentration
resulting from increased availability due to higher flowrates. The charged species would
tend to remain in the volume near the substrate because of the greater effects of the elec-
trical field.

.-
c
0
c
m
A
' r n
rn
C
.d 44
C
0
V
3 .d
2
u3
M
.d
x
:
c
0
v)
4
0
0) c
w
200
1 , " I I
-r m
W r-
4
I
0
O
N
I 1
0
N c-
0
W
c)
c- W (9
0 0
0 0
0 0
0 0
0 0
w o
0 0
0
0
0
0
c - t - w w W (9
0 0 0
m u 3 In
m m m
0 0
0 0
3 4
0
0 4
.
0
m
0
0
rl
0 0 0 0
e z 5 F l 2
hl
I
E-
OD
I
B

\
,
t
j
\
I,
v)
2
Y .
C
0
u
m
.I C
Y
C
z
d
cu
X
p:
&
a
I
I -
a 0
n 2
2
m
201
( D o
I 2 2 2
0 0 0 0
0 0 0 0
N 4 N - F
v) P - t - P -
0 0 0 0
m o a m a
a0 a J O D O D
0 0 0 0
N N N N
0 0 0 0
v) m m v )
m m m m
I 7
v) c w m
m a m 0
m Q D m m

202
CONCLUSIONS
Chemical reacticns can be initiated and sustained by a corona discharge. Boron has been
deposited on a tungsten wire substrate in a corona discharge at relatively low temperatures.
The deposition process is electrochemical in nature, the boron being deposited cathodically,
and the mechanism for the reaction is indicated to be of the form:
(a) Emission of electrons from the cathode
(b) Ionization (or activation) of hydrogen, argon, or helium, and production of
secondary electrons which in turn ionize more hydrogen, etc.
(c) Collision of ionized hydrogen, argon, or helium with boron tribromide and
exchange of energy
(d) Formation of positive boron radicals and hydrogen bromide or bromine
(e) Transfer of the positive boron radical to the cathode
(f) Electrochemical reaction of the boron radical to form an essentially amorphous
deposit of boron
The morphology of the deposit is essentially the same as the morphology of the substrate.
There is no interaction of the boron with the tungsten and there is apparently some anisotropy
of the deposit.
The corona discharge during deposition is not essentially different from a corona discharge
developed in an inert gas system, and the same criteria and properties are extant. Like a
corona discharge in an inert gas, the coefficient of electron emission and the ability of the
electrons to diffuse along the wire are most important in obtaining a uniform corona, and
thus a uniform deposit, on the substrate. Furthermore, if the sheath of positively charged
boron radicals becomes too dense. the corona is quenched and will not recover fast enough
to prevent the breakup of the corona into points.
ACKNOWLEDGEMENT
This work was supported by the U. S. Air Force Contract AF 33(615)-2130, and was based
upon an idea presented by J. B. Story. The filament cross-sections were prepared by
W. C. Coons and the continuous system was operated by Barbara Traina. Grateful acknow-
ledgement is also made to R. N. Varney, M. E. Sibert, and A. E. Hultquist for many help-
ful discussions of gas discharges in general and of corona discharges in particular.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
REFERENCES
N. A. Kapzow, IIElelstrische Vorgange in Gasen und in Vakuum, Veb Deutscher
Verlog der Wissenschaften, Berlin, 1955
S. C. Brown, "Basic Data of Plasma Physics," The Technology Press of MIT and
John Wiley and Sons, Inc., New York, 1959
E. J. Helland, The Plasma State, Reinhold Publishing Corporation, New York, 1961
L. B. Loeb, Electrical Coronas, Univ. of Calif. Press, 1965
L. Ya. Markovskii, V. I. L'vova, Yu. D. Kondrashev, Bor. Trudy Konf. Khim, Bora;
Ego Soedineii. 36-45 (1958) cf. FTD-MT-64-427, Foreign Technology Division,
A. F. Systems Command, 1965
W. V. Kotlensky andR. Schaeffer, J. Am. Chem. SOC. 80, 4517 (1958)
R. T. Holzmann and W. F. Morris, J. Chem. Phys. 29, 677 (1958)
N. R. Dibelius, J. C. Fraser, M. Kawahata, and C. D. Doyle, Chem. Engr.
Progress, 60, No. 6, 41-44 (June 1964)
S. J. Lawrence, Chem. Engr. Progress, 60, No. 6, 45-51 (June, 1964)
J. A. Coffman and W. R. Browne, Scientific American, 212, No. 6, 90-98 (June 1965)
N. A. Lange, Editor, Handbook of Chemistry, 10th Ed., 111, McCraw-Hill Book .
Company, Inc., N. Y. 1961

203
VAPOR PHASE FORMATION OF NONCRYSTALLINE FILMS
BY A MICROWAVE DISCHARGE TECHNIQUE
D. R. Secrist and J. D. Mackenzie
International Business Machines Corporation, Poughkeepsie, 1 New York
Rensselaer Polytechnic Institute, Troy, New York
ABSTRACT
Numerous variations of the microwave discharge method have been employed to
prepare thin films. In this study, non-crystalline films were deposited at low
temperatures and pressures by the vapor phase reaction of suitable compounds
with the energetic gaseous species generated in a discharge operated at 2,450
megacycles/sec. The major part of the investigation concerned the decomposition
of metal-organic compounds of the type (R),M or (RO),M, where R may
be C2H5, C3H7. etc. Amorphous oxide films of silicon, germanium, boron,
tin, and titanium were readily formed when an oxygen plasma was maintained.
Silica films were also grown via an argon plasma. These films were deposited
on metallic and non-metallic substrates positioned outside the discharge region.
Conversely. the oxidation of silicon tetrafluoride to yield a fluorsiloxane polymer
of composition SOl. SF was found to occur only within the confines of the
plasma. It is shown that the films are free from decomposition products when
deposited above a certain critical temperature, which is dependent on the reactant
flow rate. The effects of temperature, pressure, flow rate, .and hygroscopic
nature bf the substrate on the deposition rate are examined. Optical
properties are presented. Therutilization of a nitrogen or nitrogen oxide
plasma to generate nitride or oxynitride films is discussed.
INTRODUCTION
Extensive studies have been conducted with crystalline films in regard to film
structure, epitaxy, and property variations as a function of the history of a
particular deposition process. The nature of glassy films, however. has not
been investigated with the same thoroughness. This is due, in part, to the
inability of most materials to form in the glassy state. In general, the approach
to film formation has been from the vapor phase. Some methods from the vapor
include evaporation, vapor phase hydrolysis, thermal decomposition, and
"sputtering". All of these techniques will yield non-crystalline films with the
proper conditions. However, with the exception of the latter method, relatively
high temperatures are required. As a result, most films invariably
crystallize immediately after being deposited. The non-cryetalline films prepared
by these techniques are usually materials which readily form a glaea
by the conventional cooling of a melt; e. 8.. SiOz. The method of sputtering,
on the other hand, occasionally leads to the formation of uncommon non-
crystalline films when the substrates are maintained at low temperatures
In this paper. the low temperature preparation of a wide variety of non-crystalline
films is discussed utilizing a microwave discharge method as an alternative
to sputtering. With this technique, the rate of film deposition can be,
carefully controlled by regulation of the reactant flow rates. The quality of
any particular film is shown to be a sensitive function of (1) the concentration
(1) .

and type of reaction products, (2) the chemical nature of the substrate, and
(3) the reactor (substrate) .temperature.
204
APPARATUS
A schematic drawing of the main flow system employed in this work is shown
in Figure 1. A gaseous plasma, in this instance oxygen, could be maintained in
a 1" 0. D. pyrex tube when the system pressure was less than 800 microns.
The electrodeless discharge was produced with a Raytheon PGM-10 Microwave
Generator. This unit could supply 100 watts of microwave energy at a fixed
frequency of 2, 450 mc/sec. The energized species were allowed to flow into
the bell jar through a 0. 5" dia. pyrex tube. The metal-organic compounds
selected for this work were liquids with a vapor pressure of about 0. 5 mm.
near room temperature and were contained in a graduate cylinder immersed
in a constant temperature bath. The metal-organic evaporation rate (flow rate)at
any temperature could be altered by adjusting a 1 mm. dia. Teflon stopcock.
The ultimate vacuum achieved in the system was 240 microns pressure with a
gas flow rate of 1.76 cm
3
/minute. The majority of the studies were performed
inside an aluminum furnace 2 114" in dia. x 4" long. A fused silica cylinder
1 3/8" in dia. x 2 I/Z" long served as a second furnace core. The constant
temperature zones for these furnaces extended one cm. above and below the
gas inlet port with a maximum radial temperature variation of t 5OC for points
one cm. from the cylinder axes. A fused silica balance havinga sensitivity
of i cm/mg. was employed to determine the rate of film deposition. Details
of this apparatus have been previously reported(2). In order to provide a
secondary working geometry, the basic apparatus was modified as illustrated
in Figure 2. This adaption was used to study the oxidation of silicon tetrafluoride
FILM DEPOSITION AND EVALUATION
The substrates were positioned in the furnace reaction vessel with their wide
dimension perpendicular to the gas inlet tube. The general cleaning procedure
consisted of dipping the specimens into a 48% solution of hydrofluoric acid,
followed by a rinse in distilled water. During the period when the furnace temperature
was increasing, the system was purged with the gas selected to form
the plasma. In all work, the microwave generator was operated at 90% of its
rated power output.
The films were deposited onto NaCl or KBr discs to facilitate examination by
infrared transmission. Substrates of platinum, aluminum, A12O3, and fused
Si02 approximately one cm. in diameter were prepared for the rate studies.
The index of refraction and isotropic character of film specimens obtained
from the substrates and/or gas inlet tube were determined with a petrographic
microscope. Debye -Scherrer x-ray diffraction powder patterns were made to
establish whether the films were amorphous or crystalline.

BELL JAR
STANC
TO VACUUM
SYSTEM
U T 2 HELIX BALAWE
0 TELESCOR J
AWTE CYLINDER
TEMPERATURE WTH
BAY PLATE
OXYGEN ENERGlZlffi CHAMBER
TO CAPILLARY
LEAK ASSEMaY
Figure 1 : Schematic Drawing of Microwave Discharge Apparatus.
+
si F4
I, INLET
I
'h
1)
\I,
P
\
1
J
'\
I\
MICROWAVE ANTENNA
1
I OXYGEN
I
I 4- INLET
I
I 4
SUBSTRATE 1" 0. D.
PYREX TUBE
Figure 2: Modified Microwave Discharge System.

206
RESULTS AND DISCUSSION
Preliminary studies indicated that extensive water was formed as a reaction product
when the metal-organic compounds were decomposed via an oxygen discharge.
This water could be chemically incorporated into the oxide films when the substrates
were held below a certain temperature which depended on the organic
evaporation rate and the hygroscopic nature of the substrate. For'instance, with
NaCl substrates, the silica films were found to be free of water above 18OoC
with a tetraethoxysilane evaporation rate of 0. 015 cm
3
/hour. (3) The presence
of silicic acid was detected below 18OoC by infrared transmission studies. Conversely,
with a KBr substrate and similar conditions of temperature and evaporation
rate, extensive water was incorporated into the silica film.
The,majority of the films prepared in this study were formed with low deposi-
tion rates of about 20 Elminute on NaCl substrates heated to a temperature
of 200°C. The optical properties of some of the films are presented in Table 1.
TABLE 1
COMPARISON OF THE OPTICAL PROPERTIES OF
VAPOR-FORMED FILMS AND THEIR RESPECTIVE
GLASSY OR CRYSTALLINE OXIDES
Principal Infrared Index of
Mate rial Frequency (CM-l) Refraction
Si02: vapor-formed film 1, 045;800 1.458) . 002
GeOZ :
fusion-formed glass
vapor -formed film
1, 080;800
85 0
1.458: . 002
1. 5827 . 002
B203 :
fusion -formed glass
vapor -formed film
860
1,350
1. 534-1. 607
1.470i . 002
fusion-forme d glass 1,370 1.464p)
TixOy: ' vapor-formed film 950-700 b1.7
Ti02: anatase 1,200-500 >l. 7
Sn Oy:
SnljT:
vapor-formed film
cassiterite
1,425
85 0 -5 00
1.536t - . 002
31.7
All of the films were isotropic when examined with polarized light and were
amorphous byx-ray diffraction. The silica films were formed by (1) decomposing
tetraethoxysilane, (C H 0)4 Si, in either an oxygen or argon plasma or (2)
2 5
decomposing tetraethysilane in an oxygen plasma. The germania and boron
oxide films were prepared by decomposing tetraethoxygermane and triethylborate
via an oxygen discharge. The infrared absorption frequencies and refractive
indices of the latter three vapor-formed films are similar to those
of the respective fusion-formed glasses, which suggests that a random network
structure can be achieved by methods other than the conventional cooling
of the melt. The spread in the index of refraction for germania glass was
realized by slowing cooling one specimen and quenching another from 1 loO°C;
the slow-cooled specimen had the larger refractive index. With the exception

207
of the boron oxide film, the films listed in Table 1 adhered well to the NaCl substrates.
The B 0 film was immediately coated with mineral oil after removal
from the glow dfsciarge apparatus because of its extreme moisture sensitivity
and was only partially adherent.
I
The non-crystalline titanium oxide films were prepared from titanium tetra-
isopropylate while maintaining an oxygen discharge. Since titanium oxide is
not a common glass-forming system, one may inquire as to whether or not the
viscosity of such a film would be similar to that of a glass near its transition
region, e. i. 1013-1014 poises. For many undercooled sy t ms, the growth
rate can be described in terms of the reciprocal viscosit$eas:
U = AHf (Tf-T)
3TfTf ,A L % N
where u is the rate of growth (cmlsec), T is the undercooled temperature,
v\ is the viscosity (poises), Tf is the fusion temperature, A is the mean jump
distance (cm), I-$ is the latcnt heat of fusion (ergs), and N is Avogadro's number.
If this relationship is valid for a system, then an approximate estimation
of the viscosity at the crystallization temperature can be made if the rate
of growth (crystallization) is known. In this study, the titanium oxide film exhibited
birefringence (crystallized) after 45 minutes at 325OC. An approximate
rate of crystallization can be calculated for the film if it is assumed that the
particles grew from some arbitrary value, for example, 10 8 to 110 before
crystallinity was detected. The jump distance A was taken as 2 2 units. Although
Ti02 dissociates, the heat of fusion has been reported as 15. 5 kcal/mole
at a melting point of 184Ood6). The growth rate determined from the rate of
crystallization is about 3. 7 x 10-l' cmlsec. The corresponding viscosity
at 5980K is 5. 5 x 10" poises, which would classify the film as a high1 viscous
supercooled liquid since the viscosity value is slightly below 1013-10'4 poises.
Since viscosity is approximately an exponential function of temperature, it is
reasonable to conclude that at some slightly lower tem erature the viscosity
of the film is characteristic of a solid glass, e. i. 101f-1014 poises.
An amorphous tin oxide film was formed by decomposing dibutyltin-diacetate
via an oxygen plasma. Preliminary measurements indicated that the electrical
resistivity of the film near room temperature was greater than lo7 ohm-cm.
Comparing th I. R. spectra of the film and SnOZ, it is seen in Table 1 that the
main absorption mode for the amorphous film occurs at a much higher frequency
than that of the crystalline modification. This shift can probably be attributed
to a large structural variation such as a change in coordination number.
The basic apparatus shown in Figure 1 was modified as illustrated in Figure 2 in
order to study the oxidation of SiF4 via an oxygen discharge. Research grade
SiF4 gas was metered into the pyrex chamber at a rate of 1.8 cm3/minute.
Simultaneously, dry oxygen was admitted at the rate of 5 cm3/minute. The
electrodeless discharge was maintained at 800 u total pressure. Under these
conditions, a thick non-adherent film was deposited onto the walls of the pyrex

chamber within the boundaries of the plasma in about one hour. The wall temperature
in this region was approximately 100°C. The film can be described as
a translucent blue-white gel which was amorphous by x-ray diffraction and isotropic
when observed with polarized light. The index of refraction of the gel
was estimated to be considerably less than 1.400; infrared absorption bands
were evidenced at 1080, 920, and 750 cm-’. A microchemical dnalysis of the
film indicated that its composition could be represented as (SiOl. 5F)n. It is
interesting to note that the film formed only within the confines of the plasma,
which suggests that thermal and/or microwave energy are required for the
oxidativn and/or polymerization processes.
Effect of Pressure, Temperature, and Substrate - An extensive study was made
with t e silica films with respect to the mechanism and kinetics of film forma-
tion(2’. It is reasonable to suspect that the general conclusions formulated in
this work with respect to pressure, temperature, and nature of the substrate
are applicable to most of the other oxide films discussed in this manuscript.
Support for this viewpoint is also documented in regards to the incorporation
of water-into the films(3).
Basically, the effect of minor pressure fluctuations on the rate of deposition of
silica films was found to be negligible. However, above 500 microns pressure,
the rate of deposition was rapidly decreased due to (1) an increased atomic
oxygen recombination and (2) poisoning effects from the reaction products.
The films discussed in the present work were deposited near 240 u pressure.
Typical rates of deposition for the silica films ranged from 10 to 80 A/min.
by variation of (1) the organic evaporation rate or (2) the substrate temperature.
The rate of deposition was constant with time. The relationship between
logarithm of the deposition rate and reciprocal temperature is depicted
in Figure 3 for various substrates. The data have been spread to show the
temperature dependence of deposition with each substrate. With a tetraethoxy-
- silane evaporation rate of 0. 15 cm3/hour, the silica films were free of water
and/or organic inclusions (by infrared transmission studies) above 29OoC.
At higher temperatures, the rate of deposition decreased with increasing
temperature and was ascribed to a process of physical adsorption. The apparent
heat of adsorption of silica glass on the fused silica, NaC1, and platinum
substrates was calculated to be 12. 7 t 0. 2 kcal/mole. For the aluminum
or A120 substrates, the apparent heat of adsorption of silica was calculated
to be 9. 3 t 0. 3 kcal/mole. It was concluded that the hygroscopic nature of the
A1 0 surface was responsible for the lower heat of adsorption observed.
It
2 3
was postulated that a hydrated surface tends to incorporate further hydroxyl
groups into the film structure. With this reasoning, one might expect that the
hygroscopic nature of the KBr substrate discussed earlier would also lead to
a decreased heat of adsorption. In conclusion, it is postulated that the rate of
deposition of an oxide film with a moisture sensitivity similar to that of silica
is most likely affected in the same manner, e. i. physical adsorption controlled
below 410OC. The apparent heat of adsorption of the oxide films on the various
substrates, is of course, dependent on the molecular structure of the films and
the hygroscnpic nature of the substrate.

i
I,
4
I.(
Y I-
* 0
i
f
i
3 0.t
I
I I
I I .6!3 1.85
&”~’
Figure, 3: Relationship of Logarithm Silica Deposition Rate and Reciprocal
Temperature for Deposition on Various Substrates.
Other Systems - The microwave discharge technique can be employed to deposit
a variety of thin films at low temperatures. One obvious variation of the technique
utilizes gaseous plasmas other than oxygen. For instance, in the work
with silica films it was found that (C2H5O)qSi contained sufficient oxygen to
facilitate the formation of Si02 when the c mpound was decomposed with energetic
argon species. Sterling and S~ann‘~’ have recently deposited amorphous
Si3N4 films by the reaction of silane and anhydrous ammonia in an RF discharge.
Non-crystalline silicon nitride films have also been prepared at Rensselaer
Polytechnic Institute by the decomposition of ( C2H5)4Si via a nitrogen discharge.
The use of a N20 plasma to form Si20N2 poses another equally interesting
possibility. Success to date with the latter reactions has been complicated by
the ease of formation of silica. Since metal-organic compounds are now available
for a large number of metals, it would seem that the microwave discharge
technique could lead to many new and unusual films by a simple variation of
the reactant and/or plasma compositions.
i
4

21 0
References
1. R. A. Mickelson, "Electrical and Optical Properties of Amorphous Zinc
Oxide Films", Sc. D. Thesis, M. I. T, 1963.
2. D. R. Secrist and J. D. Mackenzie, "Deposition of Silica Films by the Glow
Discharge Technique", J. Electrochem. SOC. 113 (9), 914-920 (1966).
3. D. R. Secrist and J. D. Mackenzie, Incorporation of Water Into Vapor
Deposited Oxide Films", Solid State Electronics, 9, 180 (1966).
4. G. W. Morey, "The Properties of Glass", 2nd Edition, Chap. XVI, p. 370,
Reinhdd Publishing Corp., New York, 1954.
5. David Turnbull, p. 225 in Solid State Physics, Vol. 3, edited by F. Seitz
& D. Turnbull, Academic Press, .Inc., N. Y. 1956.
6. Q. Kubachewski and E. Evans, Metallurgical Thermodynamics, p. 306,
Pergamon Press, N. Y. 1958.
7. H. F. Sterling and R. C. G. Swann, "Chemical Vapor Deposition Promoted
by R. F. Discharge", Solid State Electronics 8, (8) p. 653-54 (1965).
Acknowledgements
The authors are grateful to the Pittsburgh Plate Glass Company for the support
of a fellowship. (D. R. S.).
I
A
I

211
The Reacticn 3f Oxygen with Carbonaceous Compounds
in the Electrodeless Ring Discharge
b Chester E, Gleit
Department of Chemistry, North Carolina State University, Raleigh, North Carolina, 27607
D Abstract
The electrodeless ring discharge has recently found use in chemical analysis as
a means of removing carbon from graphitic and organic compounds preceding elemental
analysis and determination of microstructure. To determine appropriate reaction conditions,
the relationship between oxidation rate and sample temperature was studied.
Samples were immersed in a plasma produced by passing molecular oqgen through a
radiofrequency field. An inductively coupled, 13.56-We generator producing 95 watts
was utilized. Gas pressure was 1.2 torr, and flow 150 cc/min. The apparent activation
energy of pure graphite was found to be 6.5 Kcal/mol in the temperature range
'' 100 to 3OO0C, and zero at higher temperatures. The activation energy of both impure
graphite and sucrose were slightly lower. These values are consistent with those of
atomic oqgen. The reaction of graphite with C02 did not occur below 1X)OC and
yielded an apparent activation energy of 3.2 Kcal/mol in the temperature range 150
to 300OC. Oxidation rate was observed to depend on electrical field configuration.
Increasing the temperature of the reaction tube's walls decreased the rate of carbon
oxidation. Trace ions, such as Fe(I1) , Se(IV), Os(1V) , and iodide, were converted
to higher oxidation states. Neither oxidation nor volatility was observed to be
strongly temperature dependent.
0 Introduction
In 1962 a method was reported for the decomposition of organic substances based
on reaction with an oxygen plasma, produced by passing molecular gas through a radiofrequency
electrodeless discharge (5). This method has found use in a variety of
chemical studies. Loss of trace elements through volatilization and diffusion is less
than that in conventiczd dry ashing. Destruction of mineral structure is reduced.
\As a small quantity of purified oxygen is the only reagent, the possibility of chemical
contamination is diminished. Applications of this method include: microincineration
of biological specimens (l3), the ashing of coal (6) and filter paper (4), and the
recovery of mineral fibers from tissue (1). Several reviews of analytical applications
! have been published (9) (14) .
In conventional ashing both reaction rate and retention of volatile components
lare strongly temperature dependent. However, disagreement exists as to the effect of
temperature in plasm ashing. In part, this is due to difficdty in applying conven-
1 tional temperature measuring techniques to solids immersed in a strong radiofrequency
"field. The question is further complicated by the fact that a single temperature
\,cannot be assigned to the system. Electrons in the low pressure plasma are not in
, thermal equilibrium with the ions and neutral species. Furthermore, the temperature
4 of the specimen and the surrounding vessel may differ considerably. In this study the
)\ effects of temperature on oxidation rate, recombination of active gaseous species,
a d volatility of inorganic reaction products are considered.
I
I
Experimental
The reaction system employed in these studies (figure 1) consisted of a 100-cm
long, 3.5-cm I .D., borosilicate cylinder (Pyrex No. 7740). Specimens were placed on
i
I

212
the principal axis 43 cm from the gas inlet. A 0.6-cm tube, C, passed diametrically
through the cylinder and held the spechen on a central projection. Sample temperature
was regulated, in part, by circulating liquids between this tube and a constant
temperature bath. Radiofrequency excitation was supplied from a 250-watt, crystal
controlled 13.56-MHz generator, described previously (3). Except as otherwise noted,
an output of 115 watts was employed. Power was transferred to the gas by means of
impedance matching network terminating in a lo-turn coil of $-inch O.D. copper tubing
tapped 2 turns from ground.
The coil, B, which had an inside diameter of 4.5 cm and
was 12.5 cm long, was coaxial with the reaction tube and placed 10 hm from the gas
inlet.
After passage through a rotometer, molecular gas entered the reaction system
through a capillary orifice, A. Pressure was monitored by a McCloud gauge attached
to side ann E, located 50 cm beyond the gas inlet. Pressure was maintained at 1.2
torr; and flow rates at 150 cc per minute, S.T.P. U.S.P. grade oxygen and instrument
grade carbon dioxide were used.
The temperature of solid specimens in the plasma was determined by means of an
infrared radiation thermometer (Infrascope Model 3-lC00, Huggins Laboratories Inc.,
Sunnyvale, Calif.) attached to a chart recorder. This device employs a lead sulfide
detector and suitable filters to permit remote measurement of 1.2 to 2.5~ radiation
emitted by the specimen. To compensate for variations in emissitivity and inhomogeneity
in the optical field, empirical calibration curves were constructed.
Themcouples were employed to determine cylinder-wall temperatures. To eliminate
interaction with the radiofrequency field, the transmitter was inactivated during
the latter measurements.
High purity graphite rods (National Carbon Co., Grade AGKSP) were employed as
standard specimens. These rods were 0.61 cm in diameter and had a cavity machined
into their base to affix them to the cooling tube. Pellets of sucrose and carbon
containing small quantities of cupric acetate were also epployed. The latter were
produced by heating and then pressing a slurry of the salt solution and m-mesh
graphite powder.
Results and Discussion
Sample Temerature: The temperature variation of oxidation rate of graphite
exposed to the oxygen plasma is shown in figure 2. Over tpe range 120 to 300OC, the
Arrhenius equation, X = Ce'Ea/RT, fits the data well and yields a value of 6.5
Kcal/mol for the apparent activation energy, Ea. Between 300 and 450°C, the highest
temerature studied, oxidation rate is not dependent on temperature. The oxidation
of graphite by molecular oxygen at high tempgrature is strongly dependent on the purity
of the graphite. To determine if a similar effecb occurs in plasma oddation, pellets
composed of pure graphite powder and inorganic salts were utilized. The results of
the addition of 0.01 _M cupric acetate are shown in figure 2. At low temperature the
change in oxidation rate of pure and impure graphite with temperature is similar.
However, the rate of oxidation of the impure graphite becomes independent of taprature
at a lower temperature. Measurements performed with specimens contabhg 1-r ,
concentrations of cupric acetate led to results which fell between the illustrated
curves.
The gas in the law pressure electrodeless discharge is chemically similar to that k
in the positive column of a low pressure arc. Wlecule-molecule and ion-molecule
collisions are frequent. The ions and neutral species are, therefore, nearly in ther- 1
mal quildbrium.
Elastic collisions between these species and electrons are less
' frequent. At low pressure electron temperatures are quite high. In the oxygen die-
Charge atodc oxygen (%) is believed to be the most abundant active species. Higher
energy states of atomic oxygen, positive and negative ions, and electronically
1
,,

2
E,
W
t-
a
Z
9
I-
a
2
X
0
I50
too
50
30
20
A A A A
1
K
SOURCE
~ ~~ ~
Fig. 1. Experimental apparatus : capillary inlet, A;
power coil, B; specimen mounting tube, C; sidearm to
manometer, D; outlet to vacuum pump, E.
1
IO
1.5 2-0 2.5
SURFACE TEMPERATURE )/r x
Fig. 2. Rate of oxidation vs. surface temperature for
graphite rods,@, and graphite pellets containing 0.01 ,M
cupric acetate,A.
I

214
states of molecular oqgen are also present.
It is instructive to compare these results with measurements of the activation
energy of graphite with atomic oxygen removed from the luminous discharge. In severearly
studies a vdue of zero was reported (2) (16). Hennig (8) observed no dependence
on temperature at a pressure of 1.0 torr and a value of 7 Kcal/mol at 10 torr. This
difference was attributed to ozone formation. In a recent detailed study of the
atomic oxygen-graphite reaction, Marsh et al. (U) reported an activation energy of
10 Kcal/mol between U and 2OO0C, which approached zero at 350°C. 'Although the rate
of oxidation in these studies was considerably lower than that obtained in the plasm,
their values are consistent with the belief that atomic oqygen is a major reactant
within the discharge.
Of a variety of specimens, only graphite exhibited an abrupt change in activation
energy near 3Oo0C. For example, the apparent activation energy in the decomposition
of sucrose is approximately 4 Xcal/mol over the entire measurement range. The independence
of oxidation rate for graphite at high temperatures is ascribed to the formation
of surface oxides. Numerous studies of graphite combustion have shown that such surface
compounds control rate at elevated temperatures in excess oxygen. As anticipated,
increasing radiofreqiiency power to the discharge does not measurablv increase the rate
of graphite oxidation at high sample temperature.
The exhaust gas of the ovgen-graphite reaction contains both carbon monoxide a d
carbon dioxide, with the latter predominating. Tn the lumi~ous discharge region these
gases react with graphite. The results of measuremepts in which carbon dioxide was
passed through the radiofrequency field and then exposed to graphite rods activated
are shown in figure 3. Between 150 and 4OOOC the Arhennius equation fits well, yielding
an apparent activation energy of 3.2 Kcal,/mol. Below 120°C less than one mg per
hour of carbon was removed. This small weight loss is attributed to ion and electron
bombardment.
W& Conditions: In a number of experiments, increasing power beyond an optimum
value led to the anomolous result of reducing ashing rate. Earlier it was noted that
placing a dry ice-acetone trap in the exhaust stream extended the length of the luminous
discharge (5). To study this effect, wall temperatures were varied while the
carbon rod was maintained at 200°C. The results of a series of measurements are shown
in figure 4.
Assuming that the decrease in oxidation rate at increased wall tempera-
ture ip due to recombination of active species at the wall, the overall activation
energy at low temperatures for this process is approximately 2 Kcal/ml.
The major sources leading to loss of active oxygen species are recombination of
atomic oxygen and ions at the'walls of the vessel. The recombination of discharged
oxygen on glass surfaces has been investigated. Linnett and Marsden (10) reported
that the recombination coefficient of atomic oxygen on clean borosilicate glass
(Pyrex) was independent of temperature. However, positive values were observed on
contaminated surfaces. In a later series of papers Greaves and Linnett (7) studied
a number of oxide surfaces, in all cases the recombination coefficient was temperature
'
dependent. For silica apparent activation energies were 1 to 13 Kcal/mol over the
temperature range 15 to 3OOOC. The exceptional nature of Pyrex was noted.
' The second major source of loss of active species results from electron-ion
recombination at the chamber walls. The electric field restricts the drift of ions
to the chamber walls. Rate of-recombination is controlled by ambipolar diffusion,
but is limited by the presence of negatively charged oqgen ions in the discharge (15).
Although diffusion is not dependent on wall temperature, a distortion of the
symmetric electrical field will enhance wall recombination, producing an inarease in
This effect is noted in t&le 1.
The presence of a 6-cm long metallic ring on the exterior wall of the
reduced oddation rate by more than ten per cent. Grounding the ring in comn with
. wall temperature and a reduction of oxidation rate.
<
/
I
1
I

'i'
100
50
IC
21 5
2.0 2.5
SURFACE TEMPERATURE, I/T i103
Fig. 3. Rate of gassification vs. temperature for graphite
rods exposed to carbon dioxide discharge.
5c
4c
3c
20
IC
1.5 2 .o 2.5 3.0 3.5
WALL TEMPERATURE, x lo3
Fig. 4. Effect of wall temperature on the oxidation
rate of graphite.

.
216
the output coil further reduced the oxidation rate. This is a local effect, in that
oxidation rate beyond the ring was not greatly reduced. Distortion of the field can
also be introduced by changing the angle between the end of the output coil and the
reactSon tube.
le&h of the luminous discharge. Similar effects have been observed in electrodeless
discharges at higher pressure (12).
Variations from axial symmetry led to reduced rates and decreased the ,
Table I I
Effect of Field Distortion on Carbon Oxidation Rate
Cbcidation Rate
Angle w/hr
Pyrex tube 00 79
Pyrex tube and ungrounded
copper ring
Pyrex tube and grounded
copper ring
00 70
00 62
mrex tube 1.0 71
Pyrex tube 3." 63 '
Effect on Nineral Constituents: It has been reported that there is no appreciable ,,
loss of a number of metal ions in plasma oxidation (4,s). Nonvolatile species include:
Na(I), Cs(II), Cu(II), Zn(II), Mn(II), Pb(II), Cd(II), Co(II), Ho(III), Er(III), Fe(III),
Cr(IIL), As(III), Sb(SII), and Mo(V1). The effect of ashing temperature, the reason
for the surprisingly low volatility, and the final oxidation dtate of the product were
not previously explored. As part of the present study, 20 to 100 mg of compounds containing
radioactive tracers were deposited on Whatman cellulose filters. After
tupxure to ashing for a sufficient period to remove the filter paper, generally 30
dnutes, the activity of the ash was measured and the oddation state of the element
determined. Specimen temperature was adjusted during ash- by altering input power.
These measurements are summarized in table 11. In general, temperature has little
effect on retention. These results and earuer observations Indicating complete retention
of metals in compounds such as arseneous chloride and metalloporphyrFns (5) are
believed to be due to competition between volatilization and oxidation to leas volatile
compounds.
Unlike the other elements studied, the highest valence oxide of o d u ,
is the most volatile. Therefore, this element is volatilized In the plasmrr
on process.
The volatility of iodide, shown in figure 5, is also consistent wlth the hypothe- {
sis of competition between volatilization and oxidation. During these measurements,
ashing was stopped at 5 minute intervals. It is Been that loss of 1131 closely foumS
ttie curve for filter paper g ssification. No loss of 1131 occurs after the filtar
paper is remved. Loss of I f 3l varied between 15 and 35 per cent. However, none Of
the residual 1131 could be preci itated with silver nitrate, indicating that the iodide
had been oxidized. When the 113 E tracer was converted to NdO3 before ashing, all Of
the iodine activitywas retained.

I-
z
W
0
w
a
a
100
80
60
40
20
215
0
0 IO 20 30
TIME (MINI
Fig. 5. Loss of sample weight,n; and loss of Il3’
tracer,O, during ashing of filter paper in an owgen
plasma.

- Tracer
Fe59
Fe59
' ~e75
se75
,8875
se75
218
Table I1
Recovery of Trace Elements from Ashed Filter Paper
~. Highest Percent in
Max. sample Percent Oxidation Highest
Compound Temperature Recovered State Oxidation State
I
123
I11
95
120
I11
100
110
250
ll0
250
ll0
250
120,
150
* Recovered after rinsing tube with HF-HN03
V
V
V
V
I
I
VI11
.VI11
The behavior of silver ion represents a different type of ashing loss. As
opposed to the other metals in this study, only 75 to 90 per cent of the Agum could
be recovered from the borosilicate sample holder by rinsing with dilute nitric acid.
To recover the remainder of the Ague? the glassware had to be repeatedly rinsed with
a warn mixture of nitric and hydrofluoric acids. gY maintaining the system at low
temperature this difficulty was ameliorated, but not eliminated.
Conclusion: A number of active species are known to be present in the luminous
electrodeless discharge. Although a sikilarity between reactions in the plasma and
those of discharged oxygen is,noted, it cannot be concluded that within the plasma
onlythe reactions of atomic owgen are significant. The present study indicates
appropriate conditions for the application of the electrodeless discharge to decompose
carbonaceous materials prior to elemental analysis. Specifically, increasing sample
temperature to 200-300OC leads to an improvement in ash- rate without appreciably
increasing volatility losses.
91
93
100
100
--
c
-
-
--
l
Acknowledgement //
, The electrical apparatus used in this study was constructed by James Breitmsier.
The assistance of Juan Wong and Daniel Silvers is appreciated. Walter Holland, Harvey
Beaudry and Robert Reinhart assisted in preliminary studies performed under a contract
between Union Carbide Chemical Corporation and the Laboratorg for Electronics.
stw was supported by grants from the North Carolina Engineering Foundation and North
C a r o b State University Professional Development Fund.
r
I
1
'
d
'
I
t

s.,
Literature Cited
Berkley, C. , Churg, J., Selikoff, I. J. , and Smith, W. E., e.
132, 48 (1965).
New York Acad.
Blackwood, J. D., and McTaggart, F. K., -. J. w., 12, 114 (1959).
Gleit, C. E. , Microchem. J. , lo, 7 (1966).
Gleit, C. E., Benson, P. A. , Holland, W. D. , and Russell, I. J., &&. m., 36,
2067 (1964).
Gleit, C. E. , and Holland, W. D., And.. Chem., 2, 1454 (1962).
Gluskoter, H. J., m, 44, 285 (1965).
Greaves, J. C., and Linnett, J. W., Trans. Farad. u., a, 1355 (1958).
Hennig, G. R. , Dienes, G. J., and Kosiba, W., Proc. Second Intern. Conf. Peaceful
Uses At. Energy, Geneva, 1, 301 (1958). -
--
HOUahan, J. R., J. Chem. FdUC., &J A@1, U97J 392 (1966).
Linnett, J. W., and Marsden, D. G. H., m. &. w.
(1956).
I
489
(Londod E. A, a,
Marsh, H., O'Hair, T. E., and Wynne-Jones, W.F.K., Trans. Farad. &., 61, 274
(1965).
Mitin, R. V., and Pryadkin, K. K., 2. =. m., z? 1205 (1965).
Thomas, R. S., J. Cell Biol. , 2, 113 (1964).
Thomas, R. S., "Techniques of Electron Microscopy: Microincineration for the Fine
Localization of Mineral Constituents".in fISubcellular Pathology" ed. by Steiner,
J. W., Morat, H. Z., and Ritchie, A. C., Harper and Row, New York (in press).
Thompson, J. B., E. m. =. (London), 22, 818 (1959).
Wrkevich, J., and Streznewski, T., Revue L'Inst. Fran. du Petrole, 3, 686 (1958).

220
THE EFFECT OF CORONA ON THE REACTION OF CARBON MONOXIDE AND STEAM
T.C. Ruppel, P.F. Mossbauer, and D. Bienstock
U.S. Department of the Interior, Bureau of Mines,
Pittsburgh Coal Research Center, Pittsburgh, Pa.
SUMMARY I
The water-gas shift reaction was chosen for study in a corona discharge with
the aim of establishing the important variables associated with the production of
hydrogen.
The reaction was carried out in a quartz Siemen's type ozonizer having an
electrode length of 12 inches and an electrode separation of 8 nun. The electrodes
consisted of fired-silver paint--one electrode coating the inside of the inner (high
voltage) tube, and the other coating the outside of the outer (ground) tube.
The reaction was studied by means of a four-factor, two-level, factorially
designed set of experiments. This consisted of input power levels of 60 and 90
watts,. pressures of .Si and 1 atmosphere, hourly space velocities of 200 and 800, and
reactor temperatures of 127 and 527OC (400 and 800'K). Although voltage was not a
factor, the voltage ranged from 3100 to 11,000 rms volts.
The prediction equation resulting from the statistical analysis of the data
indicates that more hydrogen is produced at higher pressures, temperatures, and
power inputs, and lower space velocities. Practically no hydrogen is produced in
the absence of a discharge, regardless of the values of space velocity, temperature,
pressure, or power dissipated.
The highest yield of hydrogen within the factorial region studied was 4.5 per-
cent. An extra-factorial region, which was indicated by the prediction equation to
be fruitful, was explored and 11.5 percent hydrogen was obtained.
It was found that the water-gas shift reaction in a corona discharge is
kinetically, rather than thermodynamically controlled.
INTRODUCTION
Novel techniques are being investigated at the Bureau of Mines to introduce
energy into coal and coal products in an effort to find new uses for coal. One
technique under study involves chemical reactions in a corona discharge. Initially
the gas-phase reaction of carbon monoxide and steam to produce carbon dioxide and
hydrogen will be investigated.
The immediate aim of this studv is to establish the imoortant variables associated
with the production of hydrogen in the water-gas shift reaction in the presence
of a corona discharge. A more general goal is to determine the dependence of product
yield on the important variables in a corona discharge for several gas-phase reactions'
As knowledge is gained on the effect of corona discharge in chemical reactions, mre
complex reactions involving coal and its products will be studied. 4
<
The water-gas shift reaction, which is usually carried out industrially over
an Fefl3-Cr203 catalyst at 300-500°C and 100-300 psig, has been discussed at great
length in the literature. Summary articles are a~ailable.l-~ It is almost certain-
/
'
,'
ly a surface reaction5 as opposed to the apparent homogeneous nature of the reaction ;
in the corona discharge. Experiments with the quartz Siemen's type ozonizer to be
described have shown that no reaction occurs between water and carbon monoxide at ,p
i

7 h
\ F
221
600°C and 1 atmosphere pressure in the absence of a corona. This suggests that the
reaction is homogeneous when it does occur with a corona. That is, the reactor walls
probably do not enter into the reaction by acting as a catalyst or as a third body.
The physics6-8 and chemistry9-" of corona discharge have been surveyed by
several authors. A bibliography of chemical reactio s in electrical discharges over
a thirty-year period, 1920-1950, has been compiled. 19
PROCEDURE
The corona discharge unit is shown in figure 1. The operational scheme can be
seen in the flow diagram, figure 2. Carbon monoxide from a compressed gas cylinder
(1) is passed through an activated carbon adsorption trap (5). The metered flow (7)
passes through a flow control valve (9); a second stream of gas may be blended here
if desired. The carbon monoxide at the pressure of the system passes through a
steam generator (11). The thermocouple immediately above the reflux condenser (17)
controls the steam flow rate via a time-proportionating temperature controller (not
shown). The carbon monoxide and water flow rates may be individually varied between
0.2 and 3 ft.3fhr. by adjusting the flow controller (9) and temperature controller
settings. The carbon monoxide-steam mixture then enters the Siemen's-type reactor
through preheaters (18) and (21). The reactor is enclosed in a tube furnace (23).
The product gases pass through a series of three cold traps (25, one shown) at
-8OOC. Bypass cold traps (26) are available if required. The non-condensable gases
pass through a sample train (28,29) and then through a flow meter (30) which is at
system pressure. On leaving the flow meter the non-condensable gases pass through
the system-pressure regulator (32), vacuum pump (34), and finally a wet test meter
(35). The system is designed to operate from 0.5 to 2.0 atmospheres.
The Siemen's type reactor, shown in figures 3 and 4, consists of two concentric
quartz tubes, 45 and 33 mm. OD, having a 4 mm. annular space. The electrode length
is 12 inches and there is an 8 mm. electrode-to-electrode separation. The electrodes
consist of fired-silver paint--one electrode coating the inside of the inner (high
voltage) quartz tube, and the other coating the outside of the outer (ground) tube.
Tungsten (or optionally copper) wires are attached to the outside and inside of the
reactor with two or three coatings of fired-silver paint. The capacitance of the
reactor was measured as 86-89 pfad using an impedance bridge.
The average wall temperature was used as a measure of the temperature of the
system. Three thermocouples were equally spaced vertically along the inside wall of
the reactor (fig. 2) and three along the outside wall. The inside wall thermocouples
were at the corona potential and thus were isolated from ground. A millivoltmeter
was connected in series to each of the inside wall couples, and thus was
similarly "floated". To simplify the electrically hot thermocouple circuits, no
compensating ice junction was included; the room temperature correction was added to
each millivoltmeter reading. The corona potential had a positive effect on the
millivoltmeter readings in the hot circuit. Therefore the millivoltmeters were read
before the corona potential was impressed on the system.
taken as the average of all six thermocouple measurements.
The system temperature was
The study was made according to a factorially designed set of experiments to
determine the effect of pressure, space velocity, temperature, and input electrical
power on the yield of hydrogen.
This was a four factor, two level set of experiments as shown in Table 1.
I

222
Figure 1.- Unit for studying chemical reactions in a corona,.discharge.
1
1

223
e
I'

224
Figure 3.- Siemen's type reactor.
i
I

R
L
'
-1
9mm O.D. 5mm I.D.
45mm 0.D 41mm I.D.
2mm wall with a tolerance -.2mm
,33 mm O.D. 29 mm I.D.
/
9mm O.D.
5mm I.D.
4mm annular space
Figure 4. Siernen's t:ne reactor.
!

J
4
226 (
1
Table 1.- Factorial design of the experiments
2 level, 4 factor, full factorial
Z4 = 16 experiments 4
1:1 molar mixture, H2:CO
Leve 1
Factor LOW High
Pressure, atm. 0.5 1.0
Space velocity, hr.-' 200 800
Temperature, "K 400 800
Input power, watts 60 90
I It was found that no hydrogen was produced in the absence of a discharge, no
matter what the potential impressed on the high voltage electrode. Thus Paschen's
law curves, discharge-initiating potential VS. the product of pressure and electrode
separation, were developed for the reactor. These are shown in figure 5. A second
reactor with 12 m. electrode separation was also used to obtain these curves.
Initially it was intended to factorialize secondary voltage, rather than input
power. However, it was not possible to factorialize voltage (set a high and low
level) without losing the discharge at low temperatures or drawing too high a current
at high temperatures. At high currents the arc would puncture the quartz reactor.
Maintaining a discharge in some cases and not others would introduce a very signif-
icant unfactorialized (uncontrolled) qualitative variable, discharge vs. no discharge,
into the design. While it might be possible to factorialize discharge VS. no dis-
charge, it was known that no hydrogen would be produced with no discharge, regardless
of the power level.
Input power was factorialized since the voltage and amperage required for a
chosen power level are determined by the discharge and therefore need not meet set
requirements of a factorial design.
Although the voltage was not a factor, the voltage ranged from 3100 to 11,000
rms volts.
The above factors were chosen intuitively. Other factors, such as AC frequency,
distance between reactor electrodes, and electrical current, may also have an effect.
Future experiments using a frequency of 10,000 cps are planned to detect any fre-
quency effect on hydrogen production.
The factorial design form of experbentation has the advantages of 1) obtaining
an empirical regression equation with minimum effort, 2) isolating the magnitudes of
the effects of each variable and its interaction with other variables (synergism) on
the dependent variable, 3) allowing optimization of the dependent variable within
the levels of the factors studied, and 4) suggesting the direction to be taken out-
side the region studied for further optimization.
After the desired carbon monoxide and steam flow rates were obtained and stabilized
at .?; or 1 atmosphere, the discharge was initiated. It was monitored with an
oscilloscope (Tektronix Model RM 35A), typical oscillographs of which are shown in
figure 6.
The waveforms on the left show the voltage trace above the current trace.
The area enclosed by the Lissajous figure on the right is a measure of the true Power
used by the discharge.
I
I
i
I
4
1

Figure 5.- Paschen's law curves.
2-15-67

228
W
h
3
tn
.r(
w
rn
5
w
W
>
3
U
91
M
I4
G
v)
.rl
a
W
0
rl
m
0
.rl
E
1
W

\
'?
I
The water-free product gases were analyzed by gas chromatography. However the
hydrogen yields are reported with water present.
RESULTS AND DISCUSSION
The data were analyzed according to the linear hypothesis statistical model
resulting in the following empirical prediction equation for the percent hydrogen
(water included):
I
Pct. H2 = -1.88e.762 - (1.946+0.499)P + (0.00402+0.00042)SV
+(0.00486+0.00121)T - (O.O106K).O0830)W
+(0.0047~.000791)(P*T) - (0.0000103+0.0000006)(T~SV)
+(0.000028&H).0000132)(T-W)
where P = pressure, atmospheres
SV = space velocity, hr.-1
T = temperature, "K
W = input power, watts.
An inspection of the terms in the equation, after numerical values are inserted,
shows that tempzrature has the largest effect on the production of hydrogen, with
space velocity being the least important.
The coefficient with the largest error associated with it is in the input power
term. Since the factorial levels were maintained quite closely, the inference which
can be drawn is that the yield of hydrogen tends to be least correlatable with input
power. The reverse is also true, that hydrogen production tends to correlate best
with the interaction term T-SV.
The coefficients were determined with an IEN 7090 digital computer available at
the University of Pittsburgh, using a program for a regression equation previously
developed by Bureau personnel.
According to the equation, maximum production of hydrogen, within the range
studied, was found to be at 1 atmosphere, 200 hourly space velocity, 800%, and 90
watts input. At these levels the percentage of hydrogen was 4.5, compared with the
figure 4.1 calculated by the equation. The extra-factorial region of interest suggested
by the equation would be higher pressures, temperatures, and inputs, and
lower space velocities.
Within the limitations of our equipment, and operating on a
flow basis, this was P=2, SV=lOO, T=800, and W=lOO. The hydrogen percentage in-
creased to 8.6. By stopping the gas flow, 11.6%H2 was obtained in the product.
Extrapolating an empirical equation is always dangerous. There is no guarantee
that an indicated extra-factorial region will yield good results. In the present
case, the extra-factorial region proved to be more fruitful than predicted. The
advantage of an empirical equation is the ease in which it is obtained while representing
the data well in a compact form within the region in which the data were
taken.
If one could define the reaction temperature in a corona discharge, it would be
possible to discuss the thermodynamics of the system. However it is difficult, if
not impossible, to define an equilibrium with respect to the corona reaction. While
there are methods available for obtaining measurements of the average (translational)
temperature of the gas or plasma, there is a valid question as to the "local tmperature"
at the reaction site of the activated species. However it was attempted in
this study to correlate the average wall temperature of the system with hydrogen
production and thus to define the system sufficiently for design purposes. me
prediction equation strongly indicates that this approach is satisfactory.

230
The decrease in equilibrium constant of the water-gas-shift reaction with I
temperature is shown in figure 7. As the prediction equation indicates a positive
temperature effect on the production of hydrogen, the water-gas-shift reaction in
an electrical discharge must be kinetically rather than thermodynamically con-
I
trolled. I
The yields of hydrogen shown here are less than the yields achieved industrially13
by catalytic means. What is of more interest to us is! the empirical
determination, for several reactions, of the functional dependence of the yield of
a given product on the several independent variables in a corona discharge. This
information is a prelude to future experiments where powdered coal or coal volatiles
will be treated with certain gases in a corona discharge.
CONCLUSIONS
The conclusions that can be drawn from this study are:
(1) No hydrogen is produced in the absence of a discharge, regardless of the values
of space velocity, pressure, temperature, or power dissipated.
(2) The prediction equation arising from the statistical analysis of the data in-
dicates that more hydrogen is produced at higher pressures, temperatures, and power
inputs, and lower space velocities.
(3) Within the range of variables studied, temperature has the largest relative
effect on the production of hydrogen, while space velocity has the least relative
effect.
(4) Of the four factors chosen, input power is the least correlatable with the
production of hydrogen, while the first order interaction:
velocity) is the most correlatable.
(temperature) X (space 4
(5) The water-gas-shift reaction in a corona discharge is kinetically, rather than
thermodynamically controlled.
REFERENCES I
1. Morgan, J.J. Water Gas. Ch. in Chemistry of Coal Utilization. Vol. 2, H.H.
Lowry, ed., John Wiley & Sons, Inc., New York, N.Y., 1945. i
2. Fredersdorff, C.G. von, and M.A. Elliott. Coal Gasification. Ch. in Chemistry
of Coal Utilization, Supplementary Volume, H.H. Lowry, ed., John Wiley & Sons,
Inc., New York, N.Y., 1963.
3. Blackwood, J.D. Kinetics of Carbon Gasification. Ind. Chem. 36, 55-60, 129-133,
171-175 (1960).
4. Sherwood, P.W. Water Gas Conversion.
Petroleum (London) 24, 338-340 (1961).
1
5. Reference 1, p. 1704. ,
'
6. Loeb, L.B. Electrical Coronas, Their Basic Physical Mechanisms, Univ. of Calif.
Press, Berkeley & Los Angeles, 1965, 694 pp. I
7. Brown, S.C. Basic Data of Plasma Physics, Mass. Inst. of Tech. and John WileY
& Sons, Inc., New York, N.Y., 1959, pp. 250-273. i
8. Meek, J.M., and J.D. Craggs.
1953, pp. 148-176.
Electrical Breakdown in Gases. Oxford Univ. Press, f
4
i
4
4
f
I
,
;
I
J
t

N
I
+
0"
ON
i
I
+
0
u
/
0
0
I
0
cu
-
g
-
0
0
4
-
0
0
cn
Y
0
g:
"2
a:
a
O Q
[L
W
I-
w
F
0 3
01
(Do
ul
m
a
0
s:
0
0
t
0
0
m

232
9. Glockler, G., and S.C. Lind. The Electrochemistry of Gases and Other Di- I
electrics, John Wiley & Sons, Inc., New York, N.Y., 1939.
10. Thomas, C.L., G. Egloff, and J.C. Morrell. Reactions of Hydrocarbons in Elec-
trical Discharges. Chem. Rev. 28, No. 1, 1-70 (1940).
11. Jolly, W.L. The Use of Electrical Discharges in Chemical Syntheses. Ch. in
Technique of Inorganic Chemistry, ed. by H.B. Jonassen and A. Weissberger, v. 1,
Interscience Div. of John Wiley & Sons, Inc., New York, N.Y., 1963, pp. 179-208.
12. Akerlof, G.C., and E. Wills. Bibliography of Chemical Reactions in Electric
Discharges, Aug. 1951, Office of Technical Services, Dept. of Commerce,
Washington , D . C . I
13. Christain, D.C., and P,B. Boyd, Jr. What to Look for in CO Conversion Catalysts.
Chem. Eng. 56, No. 5, 148-150 (1949).
.
/
I
<
,
d

-
233
SOME PROBLEMS OF THE KINETICS OF DISCHARGE REACTIONS
Robert Lunt
Department of Physics
University College, London
This essay is concerned with the kinetics of reactions in electrical
discharges through gases that are either effected by, or initiated by,
the charged particles present in the gas.
The principal experimental investigations on the kinetics of such reactions
which have led to the development of the statistical theory of discharge
I reaction are recalled briefly before stating spme elaborations of the formal
analysis that are now seen to be pertinent and which have not been summarized
hitherto.
The present state of the theory is then illuminated by outlining some examples
of:
(1) the achievements of the theory in providing a detailed and
quantitative Interpretation of the values found from experiment
for the empirical reaction rate coefficients considered as
functions of the discharge parameters,
(2)
the apparent inadequacy of the theory to account for the results
of experiment in two gases, and
(3) the lacunae in the experimental data hitherto available that
impede the interpretation of the data in term8 of statistical
theory: the recognition of these lacunae may pave the way to
the better design of experiment and to a =re precise uader-
standing of the mechanism of discharge reaction and of the
usefulness of electrical discharges in chemical and in
electrical engineering.
I

234
The Plasma Induced Reaction of Hydrogen Sulfide with Hydrocarbons
F. J. Vastola and W. 0. Stacy
Department of Fuel Science
The Pennsylvania State University, University Park, Pa.
Introduction I
The interaction of sulfur with hydrocarbons under a range of experimental
conditions has been investigated by various workers. Knight and co-workers1 re-
acted excited sulfur atoms, produced by 'in situ' photolysis of COS at 25OoC, with
a range of paraffinic hydrocarbons and found as primary products only the corre-
ponding mercaptans. Bryce and Hinsnelwood2 studied the reaction of sulfur vapor
with parafinnic hydrocarbons in the temperature range 320°C to 349°C and observed
that the primary products were hydrogen sulfide and unsaturated hydrocarbons.
Study of the reaction of sulfur and methane 3-7 over a catalyst at 500°C to 7OO0C
has shown the products to be carbon disulfide and hydrogen sulfide according to
the reaction
CH4 + 2S2 -+ CS 2 + 2H2S. (1)
Thomas and Strickland-Constable8 studied the interaction of sulfur and hydrocarbons
at temperatures 1200.OK to 1500'K and in the absence of a catalyst observed no carbon
disulfide formation. With a catalyst, however, the reaction proceeded yielding
carbon disulfide, hydrogen sulfide, and hydrogen; reactions similar to (1) for
methane, above 1200°K, being increasingly superseded by those similar to (2):
CH4 + S + CS2 + 2H2*
2
Reported now are some results of a study of the reactions of excited atomic
species, generated by the dissociation of hydrogen sulfide in a plasma jet, with
methane and neopentane.
9
This study was suggested by an earlier investigation of Vastola and co-workers
into the reaction between carbon and the products of water vapor microwave discharge,
where it was observed that the presence of carbon downstream inhibited the recom-
bination, to either oxygen or water, of the active atomic species. Instead, the
active oxygen reacted preferentially with carbon.
Experimental Procedure
The plasma reactor is shown in Figure 1. The reactor consists of a 10 x 0.75
inch fused quartz tube. The gas to be brought to the plasma state is introduced
tangentially upstream from the output coil of a 1.2 KW, 120 MHz RF induction heater.
Ihe spiraling motion imparted by the tangential input increases the dwell time of
the gas in the plasma zone. The hydrocarbon gases to be reacted with the plasma
dissociation products are introduced down stream from the plasma flame.
By moving
the hydrocarbon feed tube the distance between the plasma flame and the point of
hydrocarbon introduction can be varied.
TO initiate the plasma discharge argon is passed through the reactor (0.5 CU
ft/hr, 1 Atm) and a graphite rod is placed in the induction field. After the heated
graphite rod starts the discharge it is removed and the gas feed is switched to the
desired H2S-Ar mixture. Mass spectrometric gas analyses were made before and after
the reaction.
1
i

.. .
:. ,. .I , .
i ? ..
. ,
... .
. .
,235 . . . . . . 1
. .. . I >.
, . . .
. (
. a
.*. . .
'LI
0 .
H

236
Results and Discussion
When hydrogen sulfide is introduced into an argon plasma condensing elemental
sulfur produces a dense fog dovnstream from the plasma flame. As the amount of
nydrocarbon that is introduced into this region is increased the elemental sulfur
production decreases to a point where it can no longer be observed. With large
excesses of hydrocarbon the color of the plasma flame changes from blue to orange
due to back diffusion of the hydrocarbon.
Table 1 gives the reactant and product concentrations for reactions of dis-
sociated liydrogen sulfide with methane and neopentane. With the exception of run
number 3 the only gaseous products detected were hydrogen sulfide, unreacted hydro-
carbon, carbon disulfide, and hydrogen.
Table 1
Reactant and Product Analyses
Reactant Mixture Products
Composition Mole % Composition Mole %
RUXI
No. Reactants HC H2S hK HC H2S H2 CS2 C2H2 AK
1 H2S, CH4 6.4 17.6 76.5 2.2 4.5 18.5 3.8 0 70.1
2 I, 6.4 17.6 76.5 4.3 6.7 13.4 1.6 0 71.1
3 11 25.5 9.9 64.7 13.0 1.3 21.9 3.5 3.0 57.8
4 H2S, C5Hl2 15.3 14.9 70.2 11.0 0.7 18.1 3.9 0 69.7
5 11 15.3 14.9 70.2 13.4 4.8 12.9 2.8 0 70.0
The following stoichiometry would be expected for the reaction of methane:
2H2S + CH4 -+ CS2 + 4H2. (3)
In run number 3 the acetylene was produced by the back diffusion of methane into the
plasma region.
The stoichiometry €or the neopentane reaction would be
lOH2S + C5H12 -f 5cs2 + l6HzS. (4)
The reactions of neopentane were performed with a large excess of hydrocarbon with
the hope of detecting sulfur insertion compounds OK reaction intermediates. How-
ever, hydrogen and carbon disulfide were the only gaseous species produced. This
indicates that once the neopentane reacts with one sulfur atom its resistance to
further reaction is lowered.
Table 2 shows the variation of reactant decomposition with changes of hydrogen
sulfide to hydrocarbon flow ratio and plasma-hydrocarbon injection distance. From
Table 2 it can be seen that the extent of hydrogen sulfide decomposition is a
function of the amount of hydrocarbon present downstream from the plasma flame and
the distance from the flame that the hydrocarbon is introduced. the highest - decom-
Position, 95 per cent, occurring in run number 4 with an approximately ten fold excess ’
Of neopentane introduced 1/4 inch from the flame. The figures for the decomposition ’
Of hydrogen sulfide tend to be misleading, the hydrocarbon addition does not change 1
the degree of dissociation of the hydrogen sulfide but rather it reacts with the
atomic sulfur minimizing formation of hydrogen sulfide by the recombination of the I
i

237
Table 2
Experimental Variables and Xeaction Conversion Efficiencies
Flow
HC inlet
distance Conversion of decomposed
Run
- NO. Reactants
rate
(total)
*
c.f.h.
HZS/HC
from
plasma
inches
Reactant reactants to:
decomposition CS?
H? -
H2S HC H2S IHC &S + HC
% % -%- x %
1 H2S, CH4 0.14 2.75 114 72 63 65 *lo0 %lo0
2 I, 0.14 2.75 1 59 28 33 1.100 %lo0
3 ,I 0.33 0.39 114 85 43 93 36 81
4 HzS, CgH12 0.25 1.03 1/4 95 28 55. 19 46
5 I, 0.25 1.03 1 68 12 56 30 61
* cu ft/hr.
atomic hydrogen and sulfur. The data on extent of conversion to hydrogen and
carbon disulfide in Table 2 show that in runs 1 and 2 essentially all of the
carbon from the decomposed methane is converted into carbon disulfide and the
hydrogen into molecular hydrogen. In run number 3 the acetylene produced accounts
for the missing carbon and hydrogen. In runs number 4 and 5 a large fraction of
the neopentane that was decomposed was converted into a solid product.
Table 2 also shows the percentage of the sulfur, resulting from the decom-
position of the hydrogen sulfide, that is converted into carbon disulfide; the
remainder of the sulfur appears as elemental sulfur (runs 1-3) or can be incor-
porated with the carbon-hydrogen polymeric solid that is produced in the runs
us ing neopentane .
These results indicate that hydrogen sulfide can be,completely dissociated by
an electric discharge and the excited atomic hydrogen and sulfur generated can be
reacted with hydrocarbons to produce molecular hydrogen and carbon disulfide.
Under conditions which minimize back diffusion of the hydrocarbon into the electric
discharge substantially complete conversion of the hydrogen sulfide and hydrocarbon
to hydrogen and carbon disulfide can be effected.
1.
2.
3.
4.
5.
6.
7.
8.
9.
References
Knight, A. R., Strausz, 0. P. and Gunning, H. E.,J. Am. Chem. Soc. 85, 2349,
(1963).
Bryce, W. A. and Hinshelwood, C., J. Chem. Soc. 4, 3379, (1949).
Nabor, G. W. and Smith, J. M., Ind. Eng. Chem. 45, 1272, (1953).
Thacker, C. M. and Miller, E., Ind. Fng. Chem. 36, 182, (1944).
Folkins, H. O., Miller, E. and Hennig, H., Ind. Eng. Chem. 2, 2201, (1950).
Fisher, R. A. and Smith, J. M., Ind. Eng. Chem. 42, 704, (1950).
Forney, R. C. and Smith, J. H., Ind. Eng. Chem. 43, 1841, (1951).
Thomas, W. J. and Strickland-Constable, R. F., Trans. Farad. Soc. 53, 972, (1957)
Vastola, F. J., Walker, P. L., Jr. and Wightman, J. T., Carbon 1, 11, (1963).

238
The extensive literature of "active" nitrogen chemistry does not appear to
I
record any study of competition between organic substrates. This paper reports an
investigation of consumption of hydrocarbon in the competition between an olefin,
ethylene, and a paraffin, propane. Reaction probably went to completion under most,
if not all, experimental conditions.* Consumption of hydrocarbon as a result of
both primary attack by one or more components of "active" nitrogen and of secondary
attack by reactive intermediates must be considered possible.
Expcrimental
Apparatus and metnods are described more completely elsewhere, 3'4. Active
nitrogen was generated by electrodeless glow discharge supported by 2450MHz
microwaves at a total pressure of 321 Torr. Transport rates were 230cm. sec.-l,
150p mole set.-' and 1.3t0.2~ mole set.-' for linear flow, N2 flow and N(4S) flow,
respectively. !he latter was measured by nitric oxide "emission" titration.'
Hydrocarbon substrate was introduced countercurrent3 into the active nitrogen stream
at autogenous temperature (approx. 5G°C) acm. (.G87 sec.) downstream from the glow
discharge. The gas stream psssed through liquid pitrogen traps 5Gcm. (.22,sec.)
downstream from the substrate inlet. Amounts of recovered reactants were determined
by gas chromato raphy on silica &el.4 Relative activities of 14Clabeled compounds
were determined g by proportional counting of CO2 obtained by combustion of the gas
chromatographically purified compounds in 02 over CuO at 450°. Propane and labeled
ethylene were Matheson products. Cold ethylene was Phillips Research Grade.
Data
Table I smarizzs 2ercep.t consumption of ethylene consequent upon reaction of
the pure substrate and three of its mixtures with propane with six different ratios
of N(4S). Table I1 sunmarizes analogous data for propane. The data formixtures
given in'the two Tables are derived from the same sets of experiments. Table I11
summarizes apparent relative specific rates of consumption, IlkCpH4/kC3H8tt. lhese
ratios were calculated by means of e . 1, the expression which would be appropriate
c ~ ~ ~ / ~ c logC(C2H4)J ~ H ~ I ~ (C,H4)tIq10gC(C,H8)o/(C~H8)tI ,(I)
if the relative rates of consmption depended entirely on the bimolecular reaction
of each of the substrates with the same reagent. Table IV summarizes molar
radioactivities of recovered reactants or product ethane relative to that of reactant
ethylene for the reaction of equimolar mixtures of 14C-labeled ethylene and ordinary
propane.

239
Zhbie I
Eerccnt Consumption of Ztcylene
(Total Hydrocarbon) Average Percent Consumption
(N)a Pure (C&4),/ (c3H8) e
C2H4 nb 3.0 nb 110 n6 -0.33 nb
4 19 3 13 3 25' 10 -
2 24' 3 27 3 37+ 1c 51 6
1 3?+ 4 45 4 47X 16 56 9
21 3 45 2 59 1 - 730 2
112 7PX 3 87 3 93 10 90 7
, 116 9;o 2 - 93 7 84 3
a.
b.
a.
b.
Flow rate of N(4S) is l.3*O.& mole sec.-l throughout.
Number of independent determinations averaged to give the tabulated figure.
Table I1
Percent Consumption of Propane
41" 2 39 3 39 10 37 7
480 2 - 4 5 7 4 4 3
now rate of N(4S) is l.3?S10.2p mole set.-' throughout.
Number of independent determinations averaged to give the tabulated
figure.
-
I

( rot-1 -!y-roccrbon)
240
xprr2nc &ltt1\. ., 3czctlvltlcs
"kC _.
/kc Ii I'
(14 2'14 3 8
1
Sz>craco'
(CPfi4) J( C3II.S 10 I
3.c 1.c 0.35
- o.:? 1.1 -
2 (0.30) 0.33 1.2 2.9
1 (1.2) 1.5 1.7 2.3
2/3 - 2.0 - 3.5
1/2 (2.9) 4.1 5.4 5.2
1/5
(4.0) - 4.5 3.1
a. Numbers in this column are calculated from the data for pure substrates.
(Total Hydrocarbon) is t,d.cn arbitrarily as the concentration of one of the
pure hydrocarbons. Values are based on data for identical concentrations
of pure hydrocarbons.
,
bot& ii drocarbon
Ilelative 1,iolar Activities of Recovered
Reactants and Product Etiannea
2 0.9s 7 - 0.01c 1
1 0.98 4 - - /
- 0.8 1 0.0lC 1
0.3') 2 0.5 1 O.OlC 1
a. Conpared to unrcacted ethylene; hydrocarbon reactmts equbols.
b. '&e number or^ indepcndcnt ex?erimcnts.
c.
d.
Indistinguishable from the value found for,unreacted propane.
Flow rate of N(4S) was 1.z20.2p mole set.-' throughout.
P
I
I

241
Discussion
The systematic variation of the ratio IlkC a /k 'I over the range of concentration
parameters summarized in Table I11 demonstrateg thaE3r8lative consumption of
competing hydrocarbons is not determined simply by the relative rates of attack of
N(4S) (or any set of reactive species) on ethylene and propane sinc? such determination
would lead to constancy of the ratio. A similar conclusion can be drawn from the
data of Tables I and I1 by comparison of the consumption of a given substrate at a
fixed concentration upon reaction witln a fixed proportion of N(4S) in the presence and
absence of the competing substrate. Such matched points are designated in Tables I
and I1 by identicd superscripts. If the simple model of the competition from which
eq. 1 is derived were correct, ccnd Leeping in mind the similar degrees of consumption
of the two Axtrates, -.ddition of the second substrate should always substantially
suppress the consumption of the first and should have the largest effect when substrate
is in excess over reagent. In fact, with (reactant hy&ocarbon)/(N)>/l such
suppression is absent or negligible. It becomes significant only when N(4S) is in
excess. The complexity of the reaction is also indicated by the shall.ow dependence
of percent consumption of pure hydrocarbon on the ratio, (pure hydrocarbon)/(N),
particularly with propane for which a twelvefold decrease in the raeo is associated
with increase in percent consumption by a factor of only 1.65. Such behaviour is
consistent with destruction of the substrate by both attacking reagent and reactive
+itexmediates arising from the primary attack if, as the proportion of N(*S) is raised,
/these intermediates are increasingly consumed by the reagent before they can attack
the substrate. This feature is also present with ethylene but to a smaller extent.
,
Except with the largest excess of N(*S), the value of 1'kC,Hd/kC2HiI (Table 111)
increases with decrease in the ratio (CpH.+)o/(CjH~)o at constit -val;es of (total
hydrocarbon)/ (N). 'Ihis systematic change is consistent with occurrence of competition
'between the two substrates with respect to their destruction by reactive intermediates
arising from primary attack of the reagent. More specifically, (cf. Tables I and 11)
.ethylene appears to be more sensitive to consumption by intermediates arising from
propane than vice versa. Presumably, the effect of attack by reactive intermediates
can be reduced or eliminated by using sufficiently large excesses of N(*S) that the
intemediates react with the latter rather than with hydrocarbon. 'Ihis analysis
hggests that the values of 1'kC,H,/kC3He11 obtained with the largest excess of N(4S)
-most closely approximate the &e valie-of this ratio for primary attack and indicates
that this true value is at least 5. (See further discussion of the data at (hydrocarbon)
/(N) = 116 below.)
3
9
'Be above discussion tacitly asrmmes that N(4S) is the only component of "active"
\!nitrogen which is significantly involved. Ilbat this is BO for ethylene is widely
'
accepted. The work of Jones and bluer6 suggests that it is erobabl also the case
%or propane. Ilhe latter study provides a value of 1.0 x lo6 M 'set.-' for the specific
\ate of reaction of CSH~ at 50°C. Since nitrogen atom cancedtrations were determined
from HCN yields produced by reaction with ethylene, this constant should be adjusted
Uownward 30 to pk, e.g. to 7 x lo5 M-'sec.-'. Since the constant was apparently
evaluated from experiments in which propane was in molar excess over propane it may also
be significantly high because it does not correct for the consequences of attack by
,,'reactive, intermediates arising from initial attack by N(*S). Heerron has estimated from
his elegant mass spectrometric study7 of the reaction of N(*S) with ethylene that the
bate of primary attack in the (virtually temperature independent) reaction of N(4S)
(with ethylene is equal to or less than 7 x lo6 M-'=ec.-'. A retboe OS lthe ratio of
dspecific rates of primary attack by N(4S) on ethglene and prbpae, respectively, in
the vicihity of 10 is thus indicated by earlier work. Agroemmt with the present
t
I
\

242
result is well within the range of the mutual uncertainties. Herronls work further .
establishes that hydrogen atoms produced in the reaction of N(4S> with ethylene
compete with the reagent for the substrate and, in fact, provides the model upon which
interpretation of the oresent data is based.
!he data of Table IV establish that even with a sixfold excess of N(4S),
synthesis of propane4 in the reaction mixture at least in part from fragments
originating in ethylene is not detectable. Production of ethylene ejther by degradation
of propane or synthesis at least in part from fragments originating in
propane occurs to H detectable extent with a sixfold excess of N(4S) and equal initid
concentrations of the hydrocarbons. Presumably such processes are more important
with propane in excess over ethylene. !he values of 1~c3n8'1 obtained with a
sixfold excess of N(4S) are, accordingly, reduced by such replacement of consumed
ethylene, the reduction being greatest with excess propane. !his effect results in
reversal of the trend to higher values of tlkC2H4/+jHst1 with increasing excess of
N(4S) which is apparent in the data of Table 111. It can also account for the
unusual sequence of values of this ratio with increasing proportion of propane
observed with (hydrocarbon)/(N) = 6 and to a lesser extent when this ratio equals 2.
!he data of Table IV also establish that ethane is produced from both ethylene and
propane.
1.
2.
3.
4.
5.
6.
7.
Footnotes
Paper VI in the series, "Reactions of Active Nitrogen with Organic Substrates."
lhis statement is based on a study of the effect of trapping times on droduct
yields in the reaction of "active" nitrogen with propylene by Dr. Y. IPitani
in these laboratories. Cf. N. N. Lichtin, Int. J. Radiation Biol., in press.
Y. Shinocaki, R. Shaw and N. N. Lichtin, J. Am. Chem. Soc., 5 $1 (1964).
P. T. Hinde, Y. Titani and N. N. Lichtin, z., in press.
P. Harteck, G. G. Mannella and R. R. Reeves, J. Chem. Phys., 3 608 (1958).
Y. E. Jones and C. A. Winkler, Can. J. Chem., 42
&?
J. T. Herron, J. Phys. Chem., 69 2736 (1965).
d
1948 (1964).

I
\
1
243
FORMATION OF HYDROCARBONS FROM H2 + CO IN
MICROWAVE-GENERATED ELECTRODELESS DISCHARGES
Bernard D. Blaustein and Yuan C. Fu
U. S. Bureau of Mines, Pittsburgh Coal Research Center,
4800 Forbes Avenue, Pittsburgh, Pennsylvania 15213
I
Formation of hydrocarbons from H2 + CO over metal catalysts is of considerable
interest and has been studied at length. Hydrocarbons can also be formed in
electrical discharges.3, 5-9. l1-I7/ and this offers an interesting alternative for
producing hydrocarbons from H2 + CO.
Fischer and Peters?/ worked with a flow system
where the gases at 10 torr pressure circulated through a discharge (50 Hz) between
metal electrodes, then through a mercury vapor pump, a liquid air-cooled trap, and
back to the discharge. Conversion of CO to hydrocarbons was very low for one pass
through the discharge, but by recirculating the gases for 100 minutes, practically
all of the CO would react.
Unt?’ and Epple and Ap&1 worked with electrodeless radiofrequency (2-110 MHz)
discharges in static reactors at pressures up to 300 torr. Conversions of H2 + CO
and H2 + C02 to C q were quite high f?f/reaction times of several minutes; no other
hydrocarbons were formed. McTaggart,- and Vastola et a1.161 working with
electrodeless microwave (2450 MHz) discharges in flow systems at pressures of a few
torr formed only traces of hydrocarbons from H2 + CO. Here, the gas passed through
the discharge only once, and the residence time was a fraction of a second. However,
work in our laboratory has shown that in a static reactor, and with reaction times of
the order of a minute, CO can be converted in high yields to hydrocarbons in an
electrodeless microwave discharge in H2 + CO. under conditions where essentially no
polymers are formed.
EXPERIMENTAL PROCEDURES
Reactions were carried out in 11-nun ID cylindrical Vycor reactors placed in a coaxial
cavity (Ophthos Instruments, similar to type 2A described by Pehsenfeld et a1.21)
connected by a coaxial cable to a Raytheon KV-lWA CMD-10 2450 MEz generator.
For a run, the reactors were evacuated to < 3 p and filled with either 5:l Hz-CO
or various H2-CO-Ar (typically 5.1:1.0:0.4) mixtures prepared from tank gases and
stored in the vacuum system. Product analyses were made on a CEC 21-1OX mass
spectrometer. Net power into the discharge, measured with a Microwave Devices
model 725.3 meter, was approximately 36 watts, although lower power levels could be
used. Air was blown through the cavity to cool the reactor somewhat, but the ,
estimated wall temperature in the discharge was still several hundred degrees C.
Runs with argon gave the same results as in the absence of Ar, and the ratio of
total-carbon-to-argon in the product was usually a few percent lower in the product,
but did not vary by more than + 10 percent from its value in the original mixture,
indicating that only negligible amounts of polymers were formed in the reaction.
For H2 + CO mixtures without Ar it was assumed that polymer formation was negligible,
so long as reaction conditions were similar.
As a precaution against the gradual accumlation of small amounts of polymer in the
reactors, they were cleaned before each run by maintaining a discharge in O2 at
about 10 torr for 3 minutes to oxidize any carbonaceous material present. This was
repeated twice, with fresh samples of 02. A discharge in H2 (about 10 torr) was
then maintained for 3 minutes in the reactor. This was also repeated twice, with
fresh samples of H2. Since the reactions to be studied were to be carried out in
a reducing environment, this treatment with H2 was felt to be desirable after the
O2 discharge.

244
RESULTS
Figure 1 shows the results of experiments made at initial gas pressure of 12 + 1
torr in 36-cm long by 11-mm I D reactors. The percent of carbon present in the
product as C2Hp and as CH4 + C2H2 is plotted vs. reaction time. Figure 2 shows
the percent of carbon present in the product as CH4 for the same runs. Water, C02.
and occasionally slight traces of other hydrocarbons, were also present in the
product.
times of 30-120 seconds. Figure 3 shows the results of experimentslmade at initial
gas pressures of 50 t 3 torr in 20-cm long by 11-mm ID reactors, Here, the percent
of carbon present as CH4 + C2H2 is at a m3ximum of 24-25); for reaction times of
3-4 minutes. There is considerable scatter In the data; the dashed curves are
intended to show only the general trend.
The conversion of CO to CHq + C2H2 reaches a maximum of 17-18% for reaction
These data show that conversion of CO to hydrocarbons in the discharge under these
conditions is Limited. The composition of the gases in the reactor approaches a
stationary state, for reaction times of the order of 0.5 to 3 minutes, depending
upon the pressure. Because of the geometry of the reactors, and the volume of a
reactor relative to the volume occupied by the discharge, the time required to
reach the stationary state appears to be longer than is actually so, due to the
relatively long times required for diffusion of gases into and out of the discharge.
(The 36-cm long reactors were used for the runs made at 12 torr; for the runs mede
at 50 torr, the shorter 20-cm long reactors were used.
Preliminary experiments
showed that the conversion of CO to CH4 + C2H2 at 50 torr in 36-cm long reactors
was essentially the same as for the 20-cm long reactors, but took 1-2 minutes
longer to achieve mximm conversions.)
High yields of gaseous hydrocarbons from H2 + CO have now been shown to occur in
low-frequency (50 Hzll and 60 H Z / ) discharges, radiofrequency (2-110 MHz)?/ and
microwave (2450 MHz) discharges. However, in all of these cases, the reaction does
not appear to be an extremely rapid one, such as is the case for dissociation of
diatomic molecules in a discharge, for instance.
For reaction times longer than those needed to reach maximum conversions, the
conversions appear to decrease. This is probably due to some polymer fomtion,
but not enough to show up as a 10% decrease in the analytically determined C/Ar
ratio. Several runs at 12 torr (not plotted) of 5 minutes duration, and some
longer runs, including one at 50 torr for 10 minutes, gave values of the C/Ar
ratio more than 10% below the initial value of the ratio, and this was considered
evidence of polymer formation. Here, also, the yields of CH4 and C2H2 were lover.
Conversions of CO could be increased by removing reaction products ae they forrmd,
by having a cold trap surround the bottom 5 cm of the reactor before and during
the time that the discharge was maintained at a distance of 10 cm from the bottom.
The discharge is localized and extends over a distance of about 2 cm. As shown in
table 1, the conversion of CO increased markedly and the product distribution
changed. Values for the C/Ar ratio indicated that polymers were not formed. Also,
there is no indication that the product distribution is any different in the presence
of Ar than in its absence. The first three columns in the table are for the 3-
minute runs shovn in figures 1 and 2.
When the discharge was maintained for 3 minutes’
or longer, with Dry Ice (-78O) surrounding the end of the reactor, the percent Of
carbon present in the final product as CHq is 51%; C2H2, 17%; CzRg, la. At this
temperature, the only reaction product frozen out is H20. Apparently, the removal
of this is sufficient to increase the convereion of CO to hydrocarbons to approxi-
mately 78%.
I
I
/J

h
z
-
a
o m
a
20
T IME, minutes .
Fiqure 1.- Formation of hydrocarbons from H2+ CO in
microwave- generated discharge: acetylene 0,
acetylene + methane 0. Initial gas pressure =
1251 torr.
11-30-66 L-9694

20
15
IO
5
0
0
0
L 0
0
I 2 3 4
TIME, minutes
Figure2.-Formation of CH, from H,t CO in
microwave -generated discharge. Initial
gas pressure= 12 t I torr.
11-30-66 L- 9695
0
0
5
r
i

I 1 I I I I
0 I 2 3 4 5
TIME, minutes
Figure3.-For mation of hydrocarbons from He + CO in
microwave- generated discharge: methane CY,
acetylene A, acetylene + methane 0. Initial
gas pressure=50+ 3 torr.
11-30 -66 L- 9696

d
r-
0
r-
,-I
d
N
2
N
2
co
Ln
N
0
m
m
N
m
d
d
0
d
4
m
d
N
4 0
m
N
d
co
m
d
h
2
d
h
N
4
E)
d
N
0
N
QI
m
N
0
4
N
m
N
0
d
.s
0%
N
e
..
c
a
d
00
co 4
U
4
a
c
0
U
W * ?N.g
W *

The next five coluws in table 1 show that with liquid N2 cooling (-196'), the
percent of carbon in the product as CH4 in only 4%; C2H2 + C2H6 (with the latter
predominating) 81%; C3 + C4 hydrocarbons, 69.; more H20 and C02 are also formed.
The mass spectrometric analyses indicated that traces of oxygenates were also
present in the products of these runs, but in such small amunts that no identi-
fications could be attempted. For some unknown reason, C2H4 is not formed under
these conditions. (Several runs were analyzed by gas chromatography For C2H4 and
none was found.) At a temperature of -196O, the only non-condensables present in
the mixture are H2, CO, and CH4; all the other products are frozen but. One can
speculate that the high yield of C2H6 is due either to a recombination of CH3
radicals at the cold surface, or hydrogenation of C2H2, but the mechanism of
formation of any of the hydrocarbon products is not known.
The next-to-last pair of columns in table 1 show that very high conversions of CO
to C2- and higher hydrocarbons can be obtained using a 2.2:l H + CO mixture.
Here, also, the C02 production is higher, and the CH4 lower, tian compared with
the more hydrogen-rich reactant mixture. The last pair of columns, giving data
for runs made at 50 torr, indicate that C2H2 was, by far, the predominant product
in these runs.
The almost complete conversion of CO to hydrocarbonqH20, and C02, obtained by
cooling the bottom of the reactor, is reversible. Several additional experiments
(at 12 torr), where the gases were reacted for 3 minutes while the reactor was
cooled, the bottom of the reactor then warmed to room temperature in a few seconds
with a water bath, and the gases reacted for 2 more minutes, gave the same product
compositions as for the runs shown in figures 1 and 2.
Table 2 gives the results of some preliminary experiments where water vapor was
added to the H2 + CO + A r mixture before reaction. The water vapor was added to
the previously evacuated reactor and a portion of the vacuum system connected to
a Pace Engineering Co. Model P7 pressure transducer containing a + 1 psi diaphragm.
The output of this was indicated on a Pace Model CD25 transducer indicator. After
the water vapor partial pressure was measured, the stopcock to the reactor was
closed and the water frozen at,the bottom of the reactor. The reactor was then
filled with the H2-CO-Ar mixture, the bottom of the reactor warmed slightly, and
3 minutes allowed for mixing of the gases before the discharge was initiated.
Even from these few experiments, it can be seen that adding water vapor to the
initial reactant mixture has a strong inhibitory effect on the production of
hydrocarbons.
Comparison of the second and third runs listed in the table shows that 3 minutes
reaction time increases the yield of hydrocarbons only slightly as compared to'one
minute. The last two runs in the table show that when the water vapor partial
pressure is equal to or greater than the CO partial pressure in the reactant gas
mixture, no hydrocarbons are produced at all.
(The C02 yield is increased due to the reaction H20 + CO +C02 + H2.)
The pronounced inhibitory effect of H20 vapor on hydrocarbon formation in this
reaction may explain some of the scatter of the data shown in figures 1, 2, and 3.
It is possible that small (and variable) amounts of H20 vapor were adsorbed on the
walls of the reactor tubes, and that this decreased the hydrocarbon yield in (some
of) the runs by a varying amount which shows up as scatter in the data. Further
experiments would have to be done to shed more light on this point. However, there
are other, as yet unknown, sources of variability in the experimental conditions
which also undoubtedly contribute to the scatter in the data.

250
TABLE 2. - Effect of adding water to the H7-CO-Ar mixture.
Bepression of hydrocarbon formation
Run 11153
Initial H2-CO-Ar
pressure torr
(5.1 : 1.0:0.4) 10.8
Initial H20 pressure
added, torr
.35
Time of discharge, min. 1
Percent of carbon present
in product as
CH4
4.0
C2H2 4.2
co2 3.3
Hydrocarbons 8.2
DISCUSSION
11152 11154 11152
12.6 13.9 11.8
.85 .95 1.83
1 3 1
1.9 1.7 0.0
1.8 2.6 0.0
4.6 4.5 8.8
3.7 4.3 0
11013
12.7
1 6
1
0.0
0.0
23.1
0 .i
I
the reaction in I 1
a stationary i
The different conversions of CO can be explained by assuming that
the discharge. without cooling the bottom of the reactor, reaches
state where-the production of-CH4 and C2W2 is limited by-the back reaction of these
hydrocarbons with H20 and/or C02 to form H2 + CO. The data in figure 3, by comparison
with the runs in figures 1 and 2, indicate that conversion of CO to hydrocarbons is
increased at higher initial pressures, as would be expected for a reaction where the
volume of the products is less than the reactants. When one or more of the reaction
products are removed from the discharge zone by being frozen out, the Stationary
state shifts and more CO reacts to form hydrocarbons. When the frozen-out hydrocarbons
are re-introduced into the discharge, they react very readily to re-form
the initial stationary state composition. If, on the other hand, water vapor i0
added to the initial reactant gas mixture, the conversion of CO to hydrocarbons is
repressed.
These experimental observations can be sunmed up by discussing the reaction
H + CO + C q + H20 in terms of a stationary state in the discharge which can
shft in the direction that would be predicted by applying Le Chatelier's principle.
Actually, in this respect, the system behaves as if it were in chemical equilibrium.
Qualitatively, this interpretation o the data is very reasonable. Fischer and
Peters,?/ Wendt and Evans,El Lunt,?j and Epple and ApS/ have all discurrcd the
production of hydrocarbons from H2 + CO in terms of equilibria in the diocharge.
However, simple arguments shov that any discussion of equilibria in discharges is
not a straightforward one because of the absence of temperature equilibri
the various species - electrons, ions, and molecules - in the discharge.- i$-*
Manes/ has pointed out that in the statistical mechanical derivation of the
equilibrium constant in terms of the part,ition functions, the functional form Of
the equilibrium constant expression follows from the conservation of atom in the
system. This leads to the likelihood that the H2-CO-CH4-B20CO2 rysta vlll eoafom
to some equilibrium "constant" in an electrical discharge system, although th
magnitude of this constant vi11 not be derivable from equilibrium thermodynndc
properties.
1
1
,
1
)
j
I
' 1

\
.\
t
251
Given the existence of this equilibrium "constant" for (some) reactions occurring
in a discharge, we would then expect Le Chatelier's principle to apply. In fact,
the "constant" need be only approximately constant, as conditions change over a
certain range, for the system to exhibit qualitative changes in accordance with
Le Chatelier's principle.
For a preliminary inspection of the quantitative aspects of these equilibria
"constants," from the data in table 2 and the tWo one-minute runs in figures 1 and
2, values were calculated for I
and
P X P
CO H20
K1 = P XP
co2 H2
P X P
CH4 H20
K2 = ---
P x P3
GO n2
These are given in table 3. If these equilibria had been established in an ordinary
thermal system, the values for K1 and K2 would depend only on temperature. However,
these "equilibria" were established in a reactor system where part of the gas was in
an electrical discharge. Since we cannot define one temperature for the discharge,
the experimental values obtained for K1 and K2 depend on the parameters which
characterize the discharge, such as rate of energy input, geometry of the reactor,
the volume of gas in the discharge relative to the volume of the reactor, and the
electrical variables, such as field strength, etc.
The values obtained for K1 and K2 can be used to define a "chemically equivalent
temperature" for the gases in the discharge. Thus, the average value for K1
corresponds to a "temperature" of 1300'K. This is very approximate, due to the
small change of K with temperature for this reaction near 130O0K. The average
value for K2 corresponds to a "temperature" of approximately 800'K. Pursuing
much the same line of thought, Epple and Apgl calculated "equivalent temperatures"
of about 750'K from values of K1 and 850'K from values of K2 obtained from the data
on the H2-CO-CQ-H20-C02 system in their static radiofrequency reactor experiments.
The significance, if any, of these "temperatures" calculated from the discharge
data, is unknown.
TABLE 3. - Values of equilibrium "constants" calculated
from the data in table 2
"Temperature, I'
Run 11153 11151 11154 11152 11013 2154 4272 OK
K1
1.4 1.8 1.9 2.0 1.9 1.1 1.9 1300
K2
K3 x Id
21
7.4
12
17
9
37
- 1/
-
K4
3.2 2.5 3.2 -
11 -
11 -
- 11 22 33 800
11 4.5 6.3 1300
11 2.1 6.9
11 me to the large concentration of H20 vapor, no hydrocarbons were produced
in these runs.

252
One unusual finding of Epple and Apt is that CH4 was the sole hydrocarbon produced
under their coaditions. Even though CH4 was present in the discharge for minutes,
no C2H2 was produced. We have repeated their experiments (using a 400 mz
generator) and can confirm this finding, as well as their values for the amount of
CO converted to CH4 in a radiofrequency discharge. On the other hand, C2H2 is
always produced in the reaction in a microwave discharge.
this difference in products in the two cases, perhaps some significance can be found
in the approximate values of 1300°K calculated from K1 for our microwave discharge
runs as contrasted with 750'K from K1 calculated by Epple and Apt. The difference
in these "temperatures" may be another indication of what we feel intuitively -
namely, that the microwave discharge is "hotter" than the radiofrequency discharge,
because of the higher value of energy input per mole of gas in the discharge.
Values were also calculated for
and
Kg =
K4 =
P x P3
CZH2 "2
P2
CH4
2
P XP
C2H2 H20
P2 x P3
CO HZ
In attempting to explain
and are given in table 3. From the average value for K3, one can calculate a
"temperatu e" for this reaction of approximately 130OOK. The values for
K4 (K4 - K: x K ) are mrch higher than there is any reason to expect, since for
this reaction, jog K = -4.6 at 300°K, and decreases with increasing temperature.
The values calculated for K4 indicate that the reaction
is not even remotely near any sort of equilibrium in the discharge; the amount of
CZHZ formed is far greater than can be accounted for at equilibrium.
In summary, it is helpful to discuss qualitatively at least some reactions in
discharges from the point of view of stationary states or equilibria, which can
19 are also attempting to discuss
shift according to Le Chatelier's princip . Admittedly, this approach te
speculative at the moment. Other workersdischarge
reactions in terms of equilibria, and it will be of interest to see how
useful these ideas prove to be in interpreting chemical reactione in discharges.
The authors wish to thank Gus Pantages, Waldo A. Steiner, and Paul Golden for their
technical assistance; A. G. Sharkey, Jr. and Janet L. Shultr for the mass-
spectrometric analyses; and Drs. Irving Wender and Frederick Kaufman for their
valuable discussions. /
,
REFERENCES
1. Arzimovich, L. A. Elementary Plasma Physics. Blaisdell Pub. Co., New York,
N. Y., 1965, pp. 5-6.
Eck, R. V., E. R. Lippincott, M. 0. Dayhoff, and Y. T. Pratt.
August 5, 1966, pp. 628-633.
2. Scbcnce, 153,
i
1
I
/
I
(
,i

I
L
253
Epple, R. P. and C. M. Apt. The Fonnation of Methane from Synthesis Gas by
High-Frequency Radiation. Gas Operations Research Project FT-27(ext.), Am.
Gas Assoc. Catalog No. 59/OR, Am. Gas ASSOC., Inc., July 1962, 47 pp.
4. Fehsenfeld, F. C., K. M. Evenson, and H. P. Broida. Rev. Sci. Instruments,
- 36, March 1965, pp. 294-298.
5. Fischer, F. and K. Peters. Btennstoff-Chem., 2, No. 14, 1931, pp. 268-273.
6. Lefebvre, H. and M. van Overbeke. Chimie et Industrie, Special No., April
7.
1934, pp. 338-342.
Lefebvre, H. and M. van Overbeke. Compt. rend., 198, 1934, pp'. 736-738.
8.
9.
10.
11.
12.
Losanitsch and Jovitschitsch in Ch. IV of "The Conversion of Coal into Oils,"
by P. Fischer, E. Benn, Led., London, 1925
Lunt, R. W. Proc. Roy. SOC. (London), m, 1925, pp. 172-186.
Manes, M. Private Co-nication.
McTaggart, F. K. Austral. J. Chem., l7, 19.64, pp. 1182-1187.
Sahasrabudhey, R. H. and S. M. Deshpande. hoc. Indian Acad. Sci., =,
1950. pp. 317-324.
13. I Sahasrabudhey, R. H. and S. M. Deshpande.
pp. 361-368.
J. Indian Chem. SOC., 27, 1950,
14.
15.
Sahasrabudhey, R. H. and S. M. Deshpande.
pp. 377-382.
J. Indian Chem. SOC., 28, 1951,
16.
1948, pp. 366-374.
Vastola, F. J., P. L. Walker, Jr., and J. P. Wightman.
1963, pp. 11-16.
Carbon, 1, No. 1,
17. Wendt, C. L. and G. M. Evans. J. Am. Chem. SOC., 50, 1928, pp. 2610-2621.
Sahasrabudhey, R. H. and A. Kalyanasundaram. Proc. Indian Acad. Sci., z,

254
RADIOFREQUENCY ELECTRODELESS SYNTHESIS OF POLYMERS:
REACTION OF CO, N2 AND H2
John R. Hollahan
Richard P. McKeever
Research and Development Department
TRACERLAB, A Division of LFE
2030 Wright Avenue
Richmond, California 94804
Introduction
A certain amount of the large literature concerning the synthesis of compounds
in electrical discharges relates to the synthesis of polymeric materials, and many
industrial patents exist in this category. Most of these syntheses have employed
1) hydrocarbon reactant( s) or 2) discharge apparatuses involving internal electrode
configurations as the source of excitation, and the glow discharge or "cold plasma"
was initiated by a. c. or d. c. fields of several hundred volts between the internal
electrodes (1).
We wish to report an electrodeless discharge synthesis of a polymeric material
involving the simple inorganic gases CO, N2 and H2. The advantages of the electrodeless
technique are principally avoidance of electrode erosion or kinetic interaction
of the electrodes with the reacting gas mixtures, the capability of filling uniformly
a plasma reactor with a volume of excited gas, and engineering advantages
of radiofrequency excitation.
Run Procedure
Experimental
An apparatus for precision metering of several gases into plasma reactors was
designed and constructed for this work. Figure 1 shows the system design. Up to
four different gases may be metered individually at known flowmeter pressures.
Since the gases were non-corrosive, mercury manometers adeq ately served as the
flowmeter pressure measurement device. Flows of gases in cm' min-' NTP were
calculated from the gas viscosity at the operating temperature, the gas density at
the operating pressure and temperature, and the flowmeter scale reading. The flowmeter
operating pressure P was calculated from P = AP t P where Psys
is the system pressure measured
FM
at Point A and AP i%%e different%lspressure
across the manometer.
The accuracy of flow rates are taken to be f 2% to f 10% depending on flow meter
condition and flow rate. In nearly every case, the f 2'7'0 accuracy was maintained.
The reagent gases, sources, and statement of purity are as follows:
-
Gas Grade Source Purity
co c. P. Mathe s on 99. 570
N2
H
Extra Dry
Pre- Purified
II
,I
99.7%
99. 95%
Coleman
11
99. 99%
After calculations of the desired flow rates and consequent required flow conditions,
the entire gas line is purged by alternate pumping and filling with the desired
gases. The metering valve V(2) and the flow meter pressure regulating valve V(3)
are adjusted to provide the conditions of flow meter pressure P and flow meter
indication necessary to produce with accuracy the desired flow-Mes. Leak checks

1 '
Fig. 1 Schematic of flow discharge apparatus and gas monitoring system.
I
, .\\
\
255

256
with a helium mass spectrometer leak detector are made before and after the experi-
mental run. Vrlve V(1) is a gas flow on-off valve.
With the flow rates accurately adjusted, the RF Generator/RFG-600 (see below)
connected to the capacitive exciter is energized and the activator tuned while bringing
the generator to maximum power, - 300 watts. Trap T(l) which is filled with 3
mm diameter glass beads and some pyrex wool to assure efficient trapping is chill,&
with liquid nitrogen. Traps T(2) and T(3) are chilled respectively with a dry ice/2propanol
slurrey and liquid nitrogen. T(3) contains 5 mm glass beqds again to assure
efficient trapping and to assure non-interference from vapors originating from the
pump. Next, the second RF generatorlRFG-600,connected to the inductive exciter
is energized and the activator tuned while bringing the generator to maximum power,
-- 300 watts.
With the gases and flows selected for the experiment, the reaction proceeds and
the polymer is allowed to accumulate along with the other products. After running a
length of time, the generators and gas flows are turned off and the traps allowed to
warm up. The pump is then valved off and the system let up to atmosphere with Argon.
The polymer of interest is removed from the inner surface of the center tube
of T(2) and retained for analysis.
Radiofrequency Equipment
The two plasma reactors were each activated either capacitively or inductively
with RF generators (RFG-600, Tracerlab, Inc. ) with deliverable power from mini-
mum up to 300 watts. The generators consist of an RF section composed of a two
stage system utilizing a crystal controlled oscillator, and a Class C power amplifier
stage. The second part is a power supply producing plate, screen, bias, and fila-
ment power. The details of the electronic requirements for generation and trans-
mission of RF energy have been described elsewhere ( 2 ).
Energy from the generator is transmitted to the exciter plates R(1), or coil,
R(2) via a plasma activator, the circuitry of which allows maximization of the RF
energf to the gas load and minimization of the reflected power back to the generator.
A great experimental advantage of this means of power delivery and measurement is
the ability the operator realizes in being able to reproduce his discharge conditions
for the different experiments or from run to run.
The net power delivered by the
generator is a function of gas type and concentrations, which in turn are related to
system pressures and flow rates. Each time an excited gas parameter is changed,
there is a concomitant change in the intrinsic impedance of the gas load, which must
be rematched to the impedance of the secondary activator circuits. Thus,
the convenience of simply being able to achieve this matching by the power meter on
the generator, which reads forward and reflected power, is a decided advantage.
-
All experiments were carried out at a radiofrequency of 13.56 MHz*487 KHz, an FCC
approved non-interference band. The choice of inductive vs.capacitive exciters for
R(1) and R(2) was rather arbitrary, both were employed totest any significant differences
in the reactions induced. The conclusion is that the chemistry is independent
of the mode of excitation. There are, however, definite advantages either configuration
offers in certain systems, such as gas-solid reactions.
cribed in Ref. (2).
Observations and Discussion
This has been des-
I
;
Synthesis
The polymers to be described below can be produced either of two ways: (1) the
reaction directly of CO, H2, and N2; or (2) the reaction first of C02 and H2 to produce
CO, via the water gas reaction, and then subsequent reaction of the produced
1
i
,)

5'
.
'
I 1
257
co, with excess H2, and N admitted either downstream of T(1), as indicated in
2
Fig. 1, or in the manifold system. The latter experiment was believed to be a purer
The liquid nitrogen.
Source Of CO and was employed for most of the preparations.
temperature of T( 1) and its packing trapped all significant water produced in the
water-gas reaction, as well as remaining COz. It became apparent that if T(l) does
not trap all the significant water, that no polymer forms at T(2), due either to a reaction
of H20 or discharge products therefrom (3) with polymer precursors, or to
the interference of water to the deposition of the polymer as a film on the cold inner
tube of T(2). I
The polymers at low N flows are characterized by being transparent, yellow,
tightly adherent films. At fiigher N2 flows, the transparency decreases, and film
strength appears to decrease, indicating perhaps lower molecular weight materials
being formed.
The production of the polymer under the flow conditions listed in Tabie I appears
critically dependent on the system pressure. In the experiments, the flow8 bf CO or
CO and H were maintained at nearly constant values while the nitrogen flow rate
waH varied The very low flow rates of N resulted from permeation through a very
short segment of Tygon tubing initially connecting
2
R(l) to the manifold outlet. These
rates of permeation were measured by observing rates of system pressure increases
with time, and checked by He permeation with the mass spectrometer at the same
point. In Run 861, the uncertainty is probably an order of magnitude; in Runs 631
and 291, the uncertainties are half that. In any case, the low flows of N2 in the initial
experiments were much less than measured flow rates in subsequent experiments,
which indicate (Table I) that less and less polymer is produced other ases fixed) as
ca. 107 cm' min-', no polymer
was o%served to form at all.
1 the N flow rate increases. At a very high N2 flow, -
1
).
TABLE 1. Flow Conditions for Several Sample Preparations
I
Run
I Flow Rates , cm3 min-l
Power, Watts Run
~
h '
b
co2 co
N2 H2
861 1. 18 --- 1-3. 1~10-~ 3. 65
63 1 1.43 --- 3. 9x10-3 3. 95
I 291 ---- 2. 18 -42. 5x10-' 4.43
bl 931 1.25 --- 1. 26 3. 60
6 890 1.36 --- 107 3. 74
i
I 991 1.36 --- 9. 88 3. 83
psYs, Time;
torr R(1) R(2) Mins.
.310 330 1, 355 680
.320 --- 280 414
.36 300 --- 720
.45 320 350 463
.75 320 410 460
2.47 320 360 273
1 Table I1 indicates elemental analyses for C, H, N, and 0 for several runs. The per
cent nitrogen is found to increase as the flow of N increases. This suggests, up to
a certain point, that one can produce in this experiment 2 tailored polymers containing
a controllable amount of a given constituent by controlling its flow rate. The dependence
of per cent nitrogen found in the polymer on flow rate is given in Figure 2.
Why no polymer forms at high system pressures, ca. 1.0 mm Hg pressure, is
perhaps due to competing gas phase processes (4). ThFsame types of polymers no
doubt may be arrived at by substitution of ammonia for the N - H2 mixture, since
one often achieves many similar results irrespective of whetker one uses N
or NH3. Little molecular or atomic oxygen, as such, should be produced in 2 these +H2
.

Spectroscopy
Run 70 Element Found
C H 0 N
861 80.51 10.31 8. 06 0. 2
63 1 75. 0 9. 5 12. 7 2. 8
291 68. 19 9. 12 13.23 8.84
93 1 55.90 7.77 18.40 17.64
very little formed
no polymer formed
Total
I
99. 1
100. 0
99.38
39.71
The transmission infrared (Fig. 3) and internal reflectance, ATR (Fig. 4) spectra
for samples 291 and 633, respectively, show very nearly identical features for the
principal bands.
Structural moieties such as ’
0
-z-NH2, -CH2-, -CH3, C-N, C=N, and -OH
in organic compounds give the best comparative infrared spectral features to those
observed for the synthesized polymer. Spectra of polyacrylamides, polyacryloni-
triles and proteins (5) of some types, give quite similar spectra. Carbon-nitrogen
double or single bonds may exist in the synthesized pyymer, but there is no evidence
for the strong -CEN frequency at ca. 2240 - 2260 cm- (5). Nicholls and Krishna-
machari (6 ) found in microwave eTFctrodeless discharees at 2450 MHz emission
u
bands of NCO formed at cryogenic temperatures from the reaction
I
N(2P) tC0 (X Ct)+NCO
at the cold surface. Thus an NCO species may enter into the polymer as an important
structural entity in the present case.
Conclusions
Under the flow discharge conditions of this study, the RF electrodeless method
appears to offer interesting synthetic possibilities with the simple inorganic gases.
One might better understand the mechanism of origin of some polymers as a function
of simple excited molecular, atomic or free radical precursors, generated in
the discharge. The composition of the gas phase products formed in various reactions
would be worth monitoring. Some of the gas products in the present study are
being analyzed.
t
<
/
4

1 1 1
moo 1400 1300
c V'
I
IPOO
I
IIUO
I
1000
I
eo0
I
800
Fig. 3 1nfrared.transmission spectrum of polymer, Run 291.
I 4000 3500 SO00 2500 2000 1002 I600 1400 I200 1000 800 600
1 CY
!, Fig. 4 Infrared ATR spectrum of polymer, Run 633.
c
O
100
so
80
10
60
60
40
30
20
10
0

Acknowledgement
260
The authcrs are pleased to acknowledge the analytical assistance of Dr. Robert 1
Rinehart of Huffman Laboratories, of Wheatridge, Colorado. Mr. Jim Beaudry en- ,
gineered all the RF generator and activator circuitry, and for whose assistance we
are deeply indebted. Finally, technical discussions with Richard Bersin were most ~
helpful,
Literature Cited I
1. J. Goodman, J. Polymer. -- Sci. 44, 551 (1960); pertinent references to past works
are located here.
2. J. R. Hollahan, --- J. Chem. Ed., - 43, A401, A497 (1966).
3. N. Hata and P. A. Giguere, Can.' J. Chem. , 44, 869 (1966); F. Kaufman, Ann.
de Gkophys., 20 106 (1964). - --
P
-
4. B. D. Blaustein, U. S. Bureau of Mines, Pittsburgh, Pa. , personal communica-
tion, has studied a system similar to ours at higher pressure, but has observed
no polymer formation.
5. J. Schurz, H. Bayzer, and H. Sturchen, Makromol. Chem. 23, 152 (1957); I,
P. A. Wilks and M. R. Iszard, "Identification of FibersdTabrics by Internal J
Reflection Spectroscopy", paper presented at Fifteenth Mid-America Spectroscopy '
Symposium, Chicago, Illinois, June 2-5, 1964.
6. R. W. Nicholls and S. L. N. G. Krishnmachari, Can. J. Chem., 38 1952 (1960). 1 ---
1
I

i
25i
The Dlssocfatior of ToIuere VaFor in a Radiofrequency Discharge
Frank J. Dinan
Departrent of Cher.istry, Carisius Collem , Buffalo, Nev Tork
The emission spctra resulting from the excitation of toluene wpcr in electrode
discharges have teen investiqatec' by Schuler as a part of his extenelre work
dealine dth the behavior of Folecules in discharps.(l) He found that two distinctly
diffennt spectra worn emitted. One, the normal toluene emission rpectrum,
WE centered in the 2600-3000 A' naion snd was related to the forbidden benzenoid
abeorption band of toluene in the crude minor Image relationehip vhich characbriz%s
fluorescence emission. This spectrum, as it was obtained In the premnt study,
le shown In figure 1. In addition to this ultraviolet eniselon, Schuler also
observed a vfsible erission spectrum from tolusne vhich appeared In the 4300 to
5000 Ao reqlon, the details of vhich ven not published. This spctrum was
designated the "blue spectrum" and was aesipned to the benzyl radial vithout stronk,
supportine evidence.
In a subeequert study of the react1 n
of toluere vapor in a high voltage elec-
trode discharce, Kraaijveld and Watenran92f investigated the fornation of the
bibenzylmolecule fmr toluene and found that, under optimum conditions in a 2400 v
arc, LO$ of the toluene vapor could te converted to tiberzyl. The other products
formed under these conditione vere not lnveetigabd. This result, ae vel1 as those
obtained by Schuler, imply the existence of benzyl radicals as the major species
present in electrode ivduced tolwne discharges.
Yore recently, Stmitwieser and Ward conducted the first comprehensive study
vhich vas concerned vith the ranee of r ducts formed from the electrodelees
rnicrovave excitation of tolwne vapor.p3? In this study, flowina tolmne vapor in a
helium csrrier rrae stream was excjted by a 3 RMc lricrovave peneratar, The epsctrm
of the lfaht enitted vas not investigated, hovever the products were carefully
detemined. The composition of the mixture of producte obtained is sbovn in figure 2.
The mv minor amounts of diger biaryla which mm formed in this discharge
sueepsted that radical intenredfated ven not of dominant Importance. The lack of
fornation of the xylene isorera vas also interpreted as eupportlng the absence of
methyl radicals in the plasm.
Two other possibilities, the fiolecular cation and anion vere coneidemd as
likely intermediates in the rnicrovave discharge. Results obtained from labeling
experiments ruled out consideration of the cation, and it was tentatively concluded
that anion irtenrediates remained as the moat likely peesibility.
This result contrasted intereetingly vith the conclusione which had been
previously dravn regarc!inu the douinance of radjcals in toluene discharges and =st
some doubt on these conclueions. Hovever, work vhicb had been done in W. D. Cooke's
laboratory at Cornel1 University siivested that very significant differences might
be anticiptcd 'ietveen the products resulting from the excitation of organic vapors
in a microvave povend discharge ard the excitation of these vaprs vlth a radlo-
frequency source. It vas, therefore, decided to investigate both the spectre and the
prodrrets obtained vhen toltrepe vapor 1.88 excfted in a 28 Mc. mdiofrequency dis-
charge.
The apparatus used in this study in ehom in block diagram form in figure 3.
?loving tol-ve rapr vas pemd throuqh a discharge povered by an R.P. transmitter
opemtinp at 28 mgcycles and 100 watts output. The pressure of the vapor uas
maintajned constant at 0.10 to 0.15 mm. Materials fomed in the plam wm
collecbd in traps maintained at Qo and 80° and vere aubseqwntly investigated by
chromatoer%phic and smctroscopfc techniques. The liqht emitted from the discharge
VB- focumd into a scannine Fonochroaator and detected by 3 IP-28 photomultiplier
tube. The a~pllfierl output of the photomltiplier vas recorded electronically.
Dnder the conditions used in these experiments, tolucne vae converted
to products jn 12 to 1s yield vfth 75 to 8% of the toluene being recovered unchaneed.

2152
TOLUENE ULTRAVIOLET EMISSION
PRODUCTS FROM TOLUENE VAPOR
IN A 3 KMc* POWERED DISCHARGE
BENZENE 49
ETHYLBENZENE 30
PHENYLACETYLENE 8
STYRENE 3
XYLENE ISOMERS TRACE
BIBENZYL -0.5
DIPHENYLMETHANE - 0.2
81 PHENYL
Figure 2
"lo 1
i
i
- 0. I
I
4
1
(
j
J

L
!
,
c
PRODUCTS FROM A 28 MC.
DISCHARGE IN TOLUENE
PRODUCT
- YO
BENZENE 35
ETHYLBENZENE 41
6 I BE N Z'YL 16
01 PHENYL METHANE 7
BIPHENYL 2
XYLENE ISOMERS

254
A summary of the products formed in the R. F. powered dfscharge is presented
in figure L. It should Fe noted that these results am normalized to exclude
recovered tolmne, non-oondensatle cases and polymeric materials which in
combination accounted for 3 to 6 of the oripinal toluene vapor.
The formation of Substantial amounts of dimer biaryls; bibenzyl, diphenylmethane
and biphenyl is of prticulqr signifiance since, as noted above, the formation
of only trace amounts of dlmer bisryls vas observed in the microwsve
discharqe. Additional experiments -re condmted in which a helium carrier gas was
uaed, and no substantial difference in the nature and amounts of products formed was
noted. The chanp in product ratios, therefore, appears to reflect a change In
mechanism vhich results frm the use of an R. F. povered discharee.
The products formed jn this study, in particular the large amounts of biaqls,
indicate that free radicals am the major intennediatee leadlne to the formation of
products. The compounds fonned can be readily explained by the series of radical
combination and hydrown sbstraotion reactions involving benzyl, phenyl and methyl
radicals which are outlined in fieurn 5.
It should be noted that minor, but definite amounts of the three xJrlene isomers
-re fomed In the radiofrequancy discharee. The trace amounts of these isomers
which were formed from the dlssociation of toluene raper in a microwave powered discharge
hss previously been used aq an a ament against the presence of methyl
radlcals in the mlcrowave discharp. (3f As is shovn in figure 6, however, it has
been demonstrated that methyl radlcals mact vith toluene wp r with 3 hundred fold
preference for the side chafn rathe- than the rinc positions.~4) It is, therefore,
entirely co~slstent with the presence of methyl radicals In the R. F. disoharee that
msctfon should take place predominantly vith the toluene slde chain to fonn
ethylbenzene rather than with the rlng positiona to yield xylene isomers, although
the formation of small amounts of qlene should be anticipated.
Experiments in which speoifically labeled deuteriotoluene was passed through
the R.F. disoharee afforded additional experimental data which supported the
importance of radioal intermediates. The producte formed from the labeled toluene
were collected, separated by chromatoyraphio techniques, and the distribution of the
deuterium label determined by infrared and nuclear magnetic resonance epctroscopy
and mass spctmmetry.
The slde chain of the recovered ethylbenzene was found to be heavily deuterated.
The partisl pass spectrum of the recovered ethylbenzene compared to the mass
speotrm of un-deuterated raterfal is shevn in figure 7. The parent peak of the
recovemd material is five mass units higher than the corresponding peak in the
un-deuterated material, and deronstrates the incluslon of five deuterium atoms in
the molecule. The bene pak of the unlabeled ethylbenzene molecule resulte from a
P-15 cleawae correspondinp to the lam of a methyl group. In the spectrum of the
reoovemd saterial, the base peak results from P-18 cleavaae and clearly
demonotrataa the rethyl group of the side chain to be fully deuterated. This
molecule preswmbly forms from the combination of C7H5D2 radical vith a CD3 radical.
The distribution of the deuterium label in the benzyl frapent can be demonstrated
by considerlnc the WR epectrm of the recovered bibenzyl. In figure 8, the
100 VC. proton rapetic ma30nance spectrum of un-deutered bibenzyl and that of the
mcovered material ere compared. Ria important to note that the 5:2 ratio of
aromatic to sethylene protons observed in the non-deutereted material vould also be
observed with the prtjally deutemted nolecule if the deuterium label vere uniformly
distrituted. The ematly reduced methylene proton peak in the spectrum of the
remwred material, however, clearly demonetrates that the label remains localiced
fn the e!de chain of the hnayl fragment, and is not distributed throughout the
ring.
This result is entfrely consistent vlth the presence of fme radical interlpsdiatee
in the discharge slnce, as io show in fim 9, it has been demonstrated
that the un=ioniced benzyl radical does not undergo any type of rearrangement vhkb
t in the randomlzatlon of a label, such as the fornation of the tropyliw
z%a?eg By contrast, the benzyl cation doee maergo an immediate arranqement to
1ltm lon, which vould result in uniform distribution of the deuterium
labl. Since this is not observed, the benzyl catien can be ruled out as a
tr?8

'\
J
?. , .. . :
PRODUCT FORMATION REACTIONS
Figure 5
REACTION OF METHYL RADtCALS
WITH TRITIATED TOLUENE
RELATIVE RATE
CONSTANTS AT 85'
6- 0.76
m- 0.26
P-
c Ha-
I .oo
156
X.V. BEREZIN, et.a\., ZHUR. OBSCHE\ KH\W.>
3p,4093 (1960). C.A., a, 27153 b.
Figure 6

1.
266
PARTIAL MASS SPECTRA OF mn BENZWB
c?-18) -100
PRODUCT
Pigurt 7
PARTIAL NMR SPECTRA OF BIBENZYL
*ZcntQ
I I I
PRODUCT
(1 )
I
-100
4
4
,

'9
/
I
'\
"1
BENZYL CATION AND RADICAL BEHAVIOR
Figure 9
PROOUCTS FROM IODINE AND
TOLUENE VAPOR IN A2Sk DISCHARGE
PRODUCT
BLNLYL 10010E L)7
IO DOBEN ZEN€ IY
ETHYLBENZENE It
BENtEdE 13
METHYL IODIDE 6
UNIDENTIFIED 9
Figure EO

significant reaction int-ermedia te.
Additional experiments were conducted in which toluene and iodine vapors were
passe6 through the R.F. discharge together. Figure 10 shows the composition of the
products Eormed in this experiment. In .:iew of the known affinity of iodine €or free
radicals, (7) the formation of benzyl iodide, iodobenzene and methyl iodide together
with greatly reduced amounts of the previously observed products argues for the
presence of benzyl, phenyl and methyl radicals i'n the discharge. ;
It should also be noted that the visible emission spectrum which is shown in
figure 11 was observed vith the toluene discharge: This spectrum is similar to that 4
vhich Schuler had previously observed from electrode toluene discharges and assigned
to the benzyl radical.
In sumrary, the diCferent products and intermediates formed in this work as
contrasted to Streitwieser and Ward's study indicates that significant differences
are to be anticipated betveen R;F. and microwave powered discharges, and
unequivocal ly establish the importance of radical intermediates in the toluene R.F.
discharge. Addit!onally, the hazards involved in the Tenera1 ization of data
obtained in a specific type of discharge are readily apparent from these observations.
Literature Cited
1. H. Schuler and A. Michel, A. Naturforch., li)a. 495 (1955)
2. H.J. Kraaijveld and H.I. Waterman, Brenstoff-Chem., 43, 33 (1962),
C.A. x, 10627f
3. A. Strcftuicser and H.P.Ward, J. 4m. Cliem. Soc., 2, 539.. (1963).
,I
\
4.
5.
6.
I.V. Herezin, N.F. Kazankava and K. Martinek, Zhur. Obschei
Khim.. 2. 4093 (1960). C.A., 2, 27153b.
R.F. Pottie and F.F. Lossing, J. Am. Chem. SOC., 2. 2634 (1361)
P.N. Rylander, s. Meyerson and H.E. Grubb, J. Am. Chem. SOC. 2.
842 (1957).
' '
4
7. J.F. Carst and R.F. Cole, Tetrahedron Letters. w, 679 (1963).

a

270
R .ACTION OF 'bI.4LEI.C J.NI?Y?'lRITE WITH ARCFl;\TTC FT'rllRCC.4IIRDNS
UKDER TIi5 INFZUXNCE OF SILENT ZL'XTRIC ?ISC?!.RGX
A.T ATWSPHTRIC F'RZSSVRRE.
Stylianos Sifniades, David Jerolamcn, Robert Fuhrmam'
Allied Chemical Ccrporation, Central Research Laboratory, Box 309,
Yorristown, N. 3.
I
Maleic anhydride is lmoyn to form two olasses of adducts with
alkyl benzenes, depending on the conditions of excitation. A t reflux
. t-peratures and in the presence of. catalytic amounts of peroxides
adducts of type'I 2re formed (1,2). A free radical chain reaction is
assumed involving abstraction of a benzylic hydrogen. The chain length,
based on the ratio of product to' added peroxide, is 20 to 100.
R
I (R,Rf = II or Me; R"= Me,Et,i-Pr) I1 (R,Rf= H,Me,t-Ru,Cl)
4t FGOIU or somewhat higher temperatures and in the presence of
ultra-violet radiation aPducts of type I1 are formed (3 to 8). Excited
aromatic molecules or excited charge-transfer complexes of maleic
anhydride with an aromatic molecule are claimed tc be the reaction
intermediates. The addition is sensitized by benzophenone, but the
adduct can be formed in the absence of a sensitizer, although at a
much lower rate. It appears that the presence of benzophenone is. iridis-
pensable if sun light is used as the source of exciting radiation (9).
It is interesting to note that in this case only benzene forms an
adhct IT with maleic anhydride, while alkylbenzenes are only partly
Incorporated in poly-anhydride chains which are formed.
Recently formation of the adduct I1 of benzene and maleic anhydride
'under the influence of gamma radiation was reported (10). The adduct
is only a minor product ,of the reaction, corresponding to about 4% of
the maleic anhydride spent. The main product is a mixture of polyanhydrides.
These can be considered to arise through a free radical
chain similar to the one yielding adducts of type I.
It is then clear that in the system maleic anhydride-benzene (6r
alkylbenzene) two types of reactions are prevalent, one by free radical
chain yielding adduct T or poly-anhydrides, the other by means of
excited molecules or charge-transfer complexes yielding adduct II. It
was thought of interest to investigate the influence of silent or corona
discharge on this system. This type of discharge was chosen because it
is easy to maintain at atmospheric pressure.
-iharge apparatus was essentially a modif led Siemens ozonizer
Vertically mounted. It was made of Pyrex glass. A solution of sodium
chloride in glycerol circulating in the jacket constituted the ground
electrode, while a silver coating In the interior surface of the central

271
tube serve? as the high voltare electrode. The action mixture was
circulated at a hi-h rate from top Lo bottom through the reactor by
means of a custom-made membrane pump constructed of Halon (registered
mark of Allied Chemical Corooration). $n inert gas, nitrogen or helium,
was introdwed at the top of the reactor and vented at the bottom
through a condenser. The temwerature of the reaction mixture was measured
at the exit of the discharpe. Temaerature control was ensured by
regulating the temperature of the circulating glycerol solution.
I
Current at the desired frequencies was produced by an audiofrequency
generator (Heathkit ?lode1 17-71?) cou2led with a 1000 VA
custom-madeamdifier. It was raised to high voltage by means of a 25
KV transformer. Lower voltages could be obtained by acting on the outgut
of the avlifier. The power input to the system was monitorcd by
nekns of a watt-meter (Westinphouse Type PYb, specially compensated
for high frequency work) at the primary circuit of the transformer.
Eastman W.iite Label chemicals were user! I ithout purification. In
a tjpicpl experiment 29.L~ grams nf maleic anhydride in 416 ml of cumene
were kirculated for two hours in the appmatus while the discharge was
mainth'ined at a power level of 400 watts. Nitrogen was suppli?d at the
rate o'f'50 al/min. The teqperatiire of the reclction mixture was 127O C.
Tne renction mixture iuas ccole.' to 20° C. and the resiilting crystals
were filtered and washed with 50 ml of cold benzene. Additional crystals
here obtained after the mother liquor had been condensed to one third
of its initbal volume by flash evaporation. T!ie combined product was
drier! at 50 '-C. in a vacuum oven. It weigned 5.18 grams and had melting
point 255-7' C. Elementary analysis and neutralization equivalent were
consistent with formula I1 (R = H, R' = i-Pr). The infra-red spectrum
showed absence of aromatic cnaracter. Flash evaporation of the unreacted
cumene and maleic anhydride left 23.0 grams of a viscous residue which
had neutralization equivalent corresponding to a mixture of a 1:l and
2:l adduct of maleic anhydride and cumene. It was assumed that it containLd
adduct of formula I (R = R' = Me) together with products of
further condensation with maleic anhydride. Infra-red and NMR spectra
were consistent with this assumption.
Similar results were obtainPd ~laing bencene, toluene, and ethylbenzene
as substrates. Chlorobenzene and nitrobenzene failed to yield
a crystalline product. "he exwrimental conditions and the results
are summarized in table I.
Dis cu s s ion.
It is apparent from the present results that maleic anhydride
forms an adduct of type IT with benzene and alkylbenzenes under the
influence of silent discharge. The generally accepted path of formation
of IT in photochemical (3) or gamma-radiation induced (10) reaction is
% I

~ tation
272
It is reasonable to expect that the same path is followed also in 1
the present case of discharge excitation. Since the rate of formation
cf adduct 11 is independent of the concentration of maleic anhydride
(experiments 2 and 4) it is concluded that the formation of the intermediate
monoadduct is the rate-determining step. The same conclusion
has been reached in the photochemical fcrmation of adduct I1 (3,4).
Adduct I1 is not, however, the only product formed in the reaction;
a considerably larger amount oi resinous material (tabulated as adduct
"TyDe I") is also formed in all cases. This material arlses from a free
radical reaction similar to the one giving. rise to adduct I in peroxide- '
initiated reactions. The chain length of these reactions is about 20 to ;
100, as mentioned earlier. On the other hand it has been shown that the
quantum yield of adduct TI formed in photochemical reaction is about
0.1 (4). Since in the present experiments adduct IT is formed in amounts 4
about one quarter of the RInoUnt of "type I" adduct although about ten
excited molecules, or charge-transfer complexes, are necessary to produce
one molecule of TI, while one free radical will produce 20 to 100 !
molecules of T, it must be concluded that the great majority of active
species in the liquid phase in contact with the discharge are excited
molecules and not free radicals. Tn fact it can be estimated that 98%
to 99% of the active species are excited molecules. The absence of free
radicals as important reaction intermediates has also been deduced by
product analysis in the microwave glow discharge of aromatics in helium,
(11)
In the photochemical reaction benzene forms an adduct I1 at a higher
rate than the alkylbenzenes, which in turn react in relative rates indi-
cative of steric hindrance. No such effect was observed in the present 1
work, as shorn by the power yields in the table. These yields are also 4
measures of the rate of formation of adduct I1 since roughly equal power
levels of discharge were maintained in all cases.
Comparison of runs 1 and 2 shows that formation of adduct I1 is
fevored at fhe lower frequency with a correspondingly higher potential
difference applied across the electrodes. Although the electric field
strength between the dielectric surfaces is not necessarily proportional 4
to the externally measured potential difference (12), it appears that
electrons of higher average energy favor the production of adduct 11,
as evidenced by the fact that the rate of formation of I1 is higher in
a helium than in a nitrogen atmosphere (runs 4 and 5). It is known that
the average electron energy for equal field strength is quite larger in
the former gas (13). Of cource the presence of organic vapors modifies
the electron eneru distribution but it is reasonable to expect that 4
some difference still exists.
"he mechanism of energy transfer cannot be deduced with certiihty
f'rom the present data. ComDarison of yields at two temperatures (runs
2 and 3) shows higher yield at the higher temperature, a fact which may'
be interpreted as meaning that benzene vapors in the dircharqe are excited
by collision with electrons: at the higher temperature the conced
tration of benzene in the gas phare is higher. While this conclusion
may be correct it is not unambiguous because of mechanical difficultie~
at the lower temperature, arising from the fact that adduct 11 was
sparingly soluble in benzene at this temperature and quickly coated the
dielectric surfaces where it was partly decomposed. Another way of excf<
would involve bombardment of the liquid phase with electrons
generated in the discharge. This method of excitation has been shown to '
prevail in the discharge-Induced oxidation of acidified water and ferrou)
4
d
I
'

273
ion (1L.). Zxcitei-ion by collision with excited nitrcpr. or helium cannot
be an inportant 3rocess, cmsi+eri~~ tiiet the first excited state of
'neliixn contoins 19.81 EV of enert7y (15). Tollision with siich a species
~0~~1.' result in cxicnsivf f'rapentation of pn or,-anic rnoleclrlo, vrhereas
the preseyt results point to the fact that free radical initiation is
;.cry li lite6 it: this sgsten.
Table I. Cordcnseticn of maleic anhydride with ben;-ene nnd slkylben7enes
under tne influence of silent t lectric discharie.
Clar e, Trequ. Potent.Tgmn.Adduct, rms Yie d of I1 g.p.
.briber fas Kc/s* c KV C. Type 1 mmo:e/kw-hr C.
1 250 TII benz.
l?.O
26.5;
a) 15.6
i2 .s: MA, N2 3.6 17.0 74 14.5 2.77 26.2
5 380 ml benz.
17 F TU, I!€ 3.6 13.0 75 15.1 11.16 33.3 355-57
6 390 ml tolu.
13 g XA, ?e 3.6 13.0 '32 30.0 11.79 30.4 245-60
7 168 ml -tbz.
29 . w, N;2 3.6 13.0 121 25.1 7.35 36.0 260-61
3 LlC ml cme.
7-
29 g X', 12 3 .6 13.0 1?7 23.0 5.18 29.8 255-57
T ~ product F was brown.
i.bbreviations: HA, benz, tolu, etbz, cume denote correspondingly maleic
R~I hy c? r i de , b en zene , t olu en e, ethyl h en z ene and cumsne.
References.
1. b4.G. Bickford, ".S. Fisher, Y.:!. Dcllear an? C-.E. Swift, J.Am.011
. Chern.Soc. w, 251
2. 3.Y. Beavers to Rohm and. IIaas Coy 1J.S. Patent 2,692,270 (1954)
3. T). Pryce-Smith and, b.. Gilbert, J.Chem.Soc. 1965, 918
4. G.S. 3amond and l,'.:!. !lardham, ?roc.Shexn.Son963, 63
5. D. Bryce-Snith 2nd J.Y. Lod.ye, J.Chcm.Soc. xm67.5
6. ' 2. Grovehstein, Jr, q.V. Rao and J.W. Taylor, J.Am.Chem.Soc. a,
1705 (196.1)
7. O.n. Schenck and R. Steinmetz, Tetr.Letters,.lo 21, 1 (1960)
8. H.J.F. Ants and 9. Bryce-smith, J.Chem.Soc. 4791
9. D. ?ryce-Smith, 1. 'lilbert- an?< €3. Vickery, Chznd - 1962, 2060;
British Patent 986,348 (1965)
10. Z. Raciszevski, Chem.In?. s, 418
11. A. Streitwieser, Jr, an& H.R. ':ard, J.Am.Chem.Soc. 539 (1963)
12. il. Rartnikas and J.Y.E. Levi, Rev.Sci.Instr. 2, 12%*(1966)
13. L.E. Loeb, "3asic 9rocesses of "rase.ous Electronics", University
of California Press (1961)
1L. .I. Yokohatp an? S. Tsuda, Sull.Chem'.Scc.Japan, 3, 46, 1640 (1966)
15. L.B. Loeb, "Xectric:,l Coronas", Tlniversity of California Press
(1965)

274
The Polymerization of Benzene in a Radio-Frequency Discharge
David D. Neiswender
Mobil Oil Corporation, Princeton, New Jersey
INTRODUCTION
The reactions of benzene in various types of electrical
discharges have been studied by a number of researchers over a
span of nearly 70 years. Reported results vary widely. Several
early workers (1-3), using ozonizer tubes, obtained a gummy,
wax-like substance, along with hydrogen, acetylene and other light
hydrocarbon gases. One of these researchers later obtained a liquid
and a solid, both analyzing as C24H25 compounds (4). In 1930 Austin
and Black (5) found diphenyl and a solid which they suspected to be
a polyphenylene containing phenol groups. Harkins and Gans (6).
using an electrodeless discharge, got complete conversion to a red-
. brown insoluble solid analyzing for (CH)x. Davis (7). on the other
hand, obtained diphenyl, pterphenyl, a resin (CgH4)x, hydrogen,
acetylene, ethylene and light paraffins. A 1935 U.S. patent (8)
describes a discharge apparatus for preparing diphenyl from benzene.
More recently, Streitwieser and Ward (9.10) obtained a 5% conversion
of benzene in a microwave discharge, the products being low molecular
weight gases, toluene, ethylbenzene and phenylacetylene. Stille and
co-workers (ll), using a radiefrequency discharge, got a 10% conver-
sion to poly(ppheny1enes) , diphenyl, fulvene, acetylene, allene,
and methylacetylene. Vastola and Wightman (12) obtained a solid film
and concluded from the infrared spectrum of the film that no aromati-
city remained in the polymer. Jesch et a1 (13), on the other hand,
also obtained a solid film, but interpreted its infrared spectrum as
suggesting the presence of aromatic groups as well as olefinic and
acetylenic unsaturation. The wide disparity of the results certainly
indicates there is still much to be learned about the chemistry of
benzene in electrical discharges. This disparity is probably due
largely to widely varying reaction conditions. Especially important
are considerations such as the pwer dissipated in the discharge, the
pressure, etc.
The work described here is an attempt to systematize the study
of benzene reactions in radio-frequency discharges. The only products
isolated in these reactions were diphenyl, a liquid polymer and a
solid polymer. The polymers appear to be polystyrenes.

Apparatus and Procedure
275
EXPERIMENTAL
The apparatus consisted of a 3.69 MHz radio-frequency generator
capacitively coupled to a cylindrical pyrex flow reactor by means of
two external copper electrodes. An inductively coupled tank circuit
was used to match the impedances of the generator and the reactor.
An approximate measure of the pwer dissipated in the dipcharge was
determined by measuring the voltage and current during the experiments
and making a phase correction for the capacitive component of the
reactor current. A simple schematic of the apparatus is shown in
Figure 1.
The hydrocarbon or helium-hydrocarbon mixture was metered into
the reactor at various rates. Reactor pressures from 1-20 torr were
employed. In most cases the discharge established itself as soon as
the r.f. signal was applied. If it did not, it was triggered with a
Tesla coil. Products were collected in a series of traps, one at room
temperature, one at -78OC and one at -195OC.
Chemicals
Reagent grade, thiophene free benzene (Baker) and vacuum
distilled styrene (Matheson, Coleman and Bell) were used in the
discharge experiments. The helium (Matheson) had a reported minimum
purity of 99.995%.
RESULTS
Depending on the conditions employed, the reaction of benzene
in the r.f. discharge resulted in either a complete conversion to a
solid polymer or a lower conversion to a liquid polymer and diphenyl.
The solid tends to form under conditions of high power dissipation
and/or low partial pressures of benzene in the reactor. Conversely,
the liquid polymer and diphenyl result under low power dissipation
and/or high partial pressures of benzene.
The solid polymer, which can sometimes be observed leaving
the discharge zone as a fine smoke, deposits throughout the trap
system, but principally in the dry ice trap. When collected, it is
a very light, fluffy, nearly white powder which picks up a considerable
static charge upon handling. Some of the substance's physical and
chemical properties are as follows:
1. It is completely insoluble in water and in all organic
solvents which were tested.
2. It does not melt up to 435OC.
3. Thermogravimetric analysis shows that the polymer
undergoes a stegwise loss in weight, with the loss
being complete at 580OC.
4. X-ray diffraction studies show it t o be completely
amorphous.
5. The density of a pressed pellet is 1.10 g/cc.

276
Figure 1
czb
Raoio Frequency Discharge Reactor
500 watt
radio frequency
genera tor
tuning
circuit
J. reactants
products

.
277
6. A freshly prepared sam le was found to have an electron
spin density of 2 x lop7 spins/cc.
7. Its surface area ,#(nitrogen adsorption) is 42 m2/g, a
rather high value for an organic substance.
8. It is neither thermosetting nor thermoplastic.
9. It chemisorbs oxygen from the air at room temperature,
the adsorption continuing for extended peribds of time.
10. Its carbon-hydrogen ratio is 1:l.
In those experiments which did not yield the solid polymer,
the conversion of the benzene was of the order of 30%. About 5% of
the benzene was converted to diphenyl: the other 25% was converted
to a liquid polymer with an average molecular weight of 617. This
polymer, a viscous amber liquid, was not characterized as completely
as the solid, but infrared spectral data indicate it to be very
similar structurally to the solid.
Polvmer Structure
DISCUSSION
The physical properties of the solid suggest that it is a high
molecular weight, highly cross-linked polymer with an irregular
structure. The extreme insolubility precludes spectral studies
requiring solutions. It was possible to obtain an infrared spectrum
by preparing a KBr pellet containing 2% of the solid. The liquid
polymer, which was soluble in organic solvents had an infrared spectrum
very similar to the solid.
-
Of several structural possibilities considered, the one which
agrees best with the infrared data and seems the most likely from a
chemical viewpoint is a polystyrene type structure. In Figure 2, the
infrared spectrum of a reference polystyrene film (A) is compared with
the spectra of the solid (B), the liquid (C) and a solid polymer
obtained when a styrene-helium mixture was passed through the discharge
(D). The spectra are nearly identical, the only significant differences
being the OH absorptions in the 3300-3400 cm-l range and the carbonyl
absorptions at about 1700 cm-l. These bands in all three of the
spectra from the discharge-derived polymers are due to rapid oxidation
by molecular oxygen. These bands become quite pronounced if the
polymers are allowed to stand in air for a few hours.
It is believed that the principal difference between the solid
and liquid products is that the liquid is essentially a linear polymer,
while the solid is highly cross-linked. Such a highly cross-linked
polystyrene would be expected to have a lower ratio of aromatic C-H
to aliphatic C-H bonds than would a linear polymer. Comparison of
spectrum B or D with C in Figure 2 shows that the intensity of the
C-X stretching vibrations agrees with this expectation.
Because the solid and liquid seem to be structurally similar,
the NMR spectrum of the latter was examined and compared with that of
an authentic polystyrene sample. As is generally true with polymeric
materials, the resolution was quite poor and only broad bands were

h
B
C
D
2-'?
Figure 2. Infrared Spectra
-7
/I/ ' !I '!
i
T
A = Polystyrene film
B = Solid polymer from benzene
C = Liquid polymer from benzene
D = Solid polymer from styrene

observed. The polystyrene (10% solution in ccl4) spectrum simply
showed two broad peaks - one centered at 6 = 1.50 due to aliphatic
protons and one at 6 = 7.08 (with a small companion peak at 6 = 6.58)
due to aromatic protons. The spectrum of the liquid polymer (lo”/.
solution in cc14) was quite similar with the peaks appearing at
6 = 1.60 and 7.04 (no peak at 6 = 6.58). In addition, a very small
peak at 6 = 5.70 was observed. This is probably due to protons on
olefinic double bonds, and suggests either that some unspturation is
present in the polymer backbone, or that some unpolymerized vinyl
groups are present. An attempt was made to increase the resolution
of the NMR spectra by using a time averaging computer on very dilute
solutions of the polymers, but the resolution was unchanged. Though
of limited value, the NMR data do support a polystyrene type Structure.
Coupled with the infrared data, it seems quite likely that the pOlyITIerS
are both polystyrenes.
Mode of Formation of Polymer
The most likely route from benzene to polystyrene involves
several steps, the first of which is the establishment of an
equilibrium between benzene and acetylene:
This interconversion has been observed in many high energy systems
including electrical discharges. In the next step, the benzene and
acetylene, one or both of which may be in a reactive state, combine
to give styrene which then polymerizes:
The simple linear polymer thus formed, limited to small chains such
as pentamers, hexamers, etc., would explain the liquid polymer
product.
The formation of the highly cross-linked solid polymer can
be explained by postulating the formation of polyvinylbenzenes which,
when polymerized, would yield an extensive, irregular structurer
4.
Highly cross-linked polystyrene

280
Both ionic and free radical mechanisms can be written to
explain the foregoing reactions in more detail, but these would be
strictly speculative, since no definitive experimental evidence has
been obtained. Some sort of benzene ions must form as a result of
inelastic collisions between the electrons and benzene molecules in
the plasma, but whether these ions or some derivative species are
the reactive intermediates is not known. Recently, Potter et a1 (14)
have shwn that styrene, when perfectly dry, does polymerize via an
ionic mechanism when irradiated with gamma rays. One cah picture a
similar mechanism occurring in the electrical discharge.
On the other hand, the fact that the polymer has a high
electron spin density suggests free radical involvement. However,
it can be questioned whether the unpaired electrons arose during or
after the polymerization reaction. An attempt was made to induce a
spin signal in a finely divided polystyrene sample by passing the
solid through a helium discharge. No signal was detected after this
treatment.
I
The question of mechanism must await the results of more
fundamental studies of the phenomena occurring within the discharge.
Effect of Reaction Variables
In an attempt to find a correlation between the reaction
conditions and the type of polymer obtained, about 50 experiments
were examined. The dependence of the nature of the product and the
benzene conversion on r.f. power dissipated per mole of benzene was
tested first. Those experiments which gave a complete conversion to
solid polymer had an average dose of 1.9 x LO7 watt-sec/mole 8H.
Those which gave a low conversion to liquid polymer plus diphenyl had
an average dose of only 7.0 x 106 watt-sec/mole 8H. This correlation
with dose is reasonable since more energy would be required to
furnish the additional acetylene required for cross-linking and to
gain the complete conversion.
A number of experiments within the group examined were conducted
with varying proportions of helium, neon and argon mixed with the
benzene. The presence of these gases did not alter the correlation
with dose provided that the r.f. paver dissipated was considered as
deposited in the benzene alone. This is a reasonable presumption
because the ionization potential of these gases is well above that
of benzene.
Within the group of experiments, some produced solid polymer
at doses as low as 4.9 x 106 watt-sec/mole 8H, but with reduced
conversions. This suggests that while the dose does correlate with
the conversion obtained, it alone does not determine the product.
It is evident that this system will require additional investigation
to understand how the products form.

i
i (12)
c
(11)
(14)
I 281
LITERATURE CITED
Losanitsch, S. M. and Jovitschitsch, M. Z., Ber. '2, 135 (1897)
DeHemptinne, A., Bull. sci. acad. roy. Belg. (3) 34, 269 (1897)-
DeHemptinne, A,, 2. physik. Chem. 22, 360 (1897)i 25, 284 (1898).
Losanitsch, S. M., Ber. 41, 2683 (1908).
Austin, J. B. and Black, I. A., J. Am. Chem. SOC. 52, 4552
(1930).
Harkins, W. D. and Gans, D. M. ibid. 52, 5165 (1930).
Davis, A. P., J. Phys. Chem. 35, 3330 (1931).
Kleinschniidt, R. V., U.S. 2,023,637 (1930).
Streitwieser, A. and Ward, H. R., J. Am, Chem. SOC. 84, 1066
(1962).
Streitwieser, A. and Ward, H. R., ibid. 85, 539 (1963).
-
Stille, J. K., Sung, R. L. and Vander Kooi, J., J. Org. Chem.
30, 3116 (1965).
Vastola, F. J. and Wightman, J. P., J. Appl. Chem. 2, 69
(1964) .
Jesch, K., Bloor, J. E. and Kronick, P. L., J. Polymer Sci.
- 4, 1487 (1966).
Potter, R. C., Bretton, R. H., Metz, D. J., ibid. 4, 2295
(1966)

INTRODUCTION
Chemistrv of Electrical %?scharge Polymerizations
Dr. Peter M. Ray
Central Research Laboratories, J. P. Stevens c Co.
Garfield, N. J. 1
The conversion of volatile organic compounds into liquid and solid prod- I
ucts by the action of a high-voltage gas discharge has been investigated
by many people over the past 100 years and more (1). Recently some
companies have been reported to be working on the application of this
principle to the coating of containers (2). steel strip (3) or fabric
(4). These references all have indicated that work was to be under
low-pressure conditions where the only gases would be the volatile
monomer. In this paper I will describe some results obtained under
atmospheric-pressure conditions, with the electrical discharge taking
place in a mixture of nitrogen and volatile organic compounds.
EXPERIMENTAL METHODS
All polymerizations were carried out at room temperature in a mixture
of organic vapors and nitrogen at a total pressure of one atmosphere.
The gases were delivered to the reaction zone through the system shown
schematically in Figure 1. Nitrogen passed through liquid monomer in
one bubble tube and through additive in another tube. The nitrogen
and entrained vapors entered the enclosed reaction chamber at one
corner and exited to the atmosphere at the opposite corner. The ratio
of monomer to additive was determined by weighing the tubes before and
after the experiment.
Details of the reaction chamber are shown in Figure 2. Polymeriza-
tion was initiated by corona discharge between two cylindrical,
parallel, insulated electrodes (A) made by lining the inner surfaces
of Pyrex glass tubing with aluminum foil. The glass tubing had a wall
thickness of 2.5 mm. and an outside diameter of 63 mm. The two glass
surfaces were separated by a gap of 4 nun. The alternating current
high voltage was supplied by a Tesla generator, manufactured by Lepel
High Frequency Laboratories, Inc. (Model HFSG-2). The peak voltage,
as estimated by spark length in air, was about 20,000 volts.
A moving strip of flexible substrate, (B) was positioned in the elec-
trode gap. As shown in Figure 2, this substrate, which in most cases
was 2-mil (50 microns) poly(ethy1ene terephthalate) film, was formed
into a closed loop 47 cm. long and 5 an. wide and was moved by the
rotating roller, C. Vapors entered through D and exited through E.
Polymer which formed in the corona zone deposited both on the glass
electrode coverings and on the moving substrates. The substrate strip
was dried and weighed before and after the polymerization. The coated
strips were also baked in a circulating air oven and reweighed. The
final weight gain was taken as the yield. In some cases polymer was
!

FLOW-
METER
I
ADDITIVE
SUPPLY
* ,283
VACUUM
PUMP
1
7
REACTION CHAMBER
VAPOR POLYMERIZATION - SCHEMATIC LAYOUT
Figure 1
PRESSURE
GAUGE
VAPOR POLYMERIZATION ON MOVING SUBSTRATE
Figure 2
TO HIGH
VOLTAGE

254
removed from the glass surfaces by solvent and recovered for infrared
or chemical analysis.
RESULTS
Survey of Polymerizable compounds
A number of volatile orqanic compounds were subjected to corona discharge
with a wide variety of results. In almost all cases a deposit
of oily or solid brown material formed on the substrate strip.
weight gain resulting from 15 minutes of .corona polymerization ranged
from tenths of a milligram to over 30 milligrams. When the coated
strips were heated at 15OoC for 10 minutes,, part of the added weight
was lost. Some'of this is thought to be unreacted monomer which was
absorbed by the substrate and coating. Another part of it could be
very-low-molecular-weight products of the corona reaction. A third
possibility is thermal decomposition of the coating.
The compounds which gave the heaviest coatings after heating, (taking
into account the amount of monomer volatilized) included triallylamine,
acrylonitrile, toluene and styrene. As the list in Table 1 shows it
is not necessary for the monomer to be a vinyl compound in the strjct
sense of the word in order for a non-volatile polymer product to form
in corona discharge. Toluene, benzene, benzotrifluoride and even
acetone gave measurable yields. There seems to be no pattern'of
relationship between the structure of a monomer and its yield in
corona polymerization.
Triallylamine
Acrylonitrile
To 1 ue n e
Styrene
Acrylic Acid
Benzene
I
TABLE 1
v
Benzotrifluoride
4-Vinyl cyclohexene
Ethyl acrylate
1-Octene
Allyl amine
Vinyl acetate
Acetone
The

285
TABLE 2
Effect of Haloqenated Additives on Corona Polymerization of Styrene
Yield in Milliqrams per Gram of Styrene
Additive Percent Added Yield Additive Percent Added Yield
Chloroform
Bromo f o m
0.5
0.5
22.4
11.6
Chlorine
Bromine
7
1
1
5.6
19.0
Bromoform 1.0 14.2 Bromine 5 32.0
Bromo form 2.5 19.7 Bromine 10 16.7
Bromo form 5.0 19.2 Iodine 5 9.7
Bromoform
Iodoform .
10.0
5.0
12.0
21.2
None
-
0
-
10.7
-
Other additives that enhanced the yields of styrene polymer were
carbon tetrachloride, 1,2-dibromo-1,1,2,2-tetrafluoroethane, l-bromo-
butane and 2-bromobutane. Yields from monomers other than styrene
were not all increased by halogenated additives: some even were de-
creased. It was impossible to develop any rational relationship
between monomer structure and susceptibility of the monomer to yield
enhancement by halogenated additive.
The results given above were all derived from experiments in which the
additive and the monomer were mixed and volatilized from a single
bubble tube. Thus, the exact composition of the vapor was not known.
A new set of experiments was carried out using separate bubble tubes
for_monomer and additive. The weight changes of the tubes during a
run were used to calculate mole ratios of additive to styrene. The
results of experiments with four additives are shown in Figure 3.
The conversion to polymer of styrene without additives was 0.75 to
0.9 percent. As increasing amounts of bromoform, 1-bromobutane or
2-bromobutane were added, the conversion increased and then fell off
again. The pattern of points in the case of 1-bromobutane was widely
scattered but most of the points lay well above the level for styrene
itself. With 2-bromo-2-methylpropane as an additive there was no
significant increase in conversion. The weight gain from the additives
alone, with no styrene, were all low compared to styrene.
In considering chemical explanations for corona polymerization, both
free radical and ionic intermediates are possibilities. Experiments
were run with various additives to styrene that might be expected to
inhibit each kind of reaction through combination with the active
intermediate but no clear-cut reduction in yield was observed. Benzo-
quinone at l and 2 mole percent gave normal yields. Water and butyl
amine were extensively studied but, as Figure 4 demonstrates, the
tendency for these supposed cation scavengers to depress the conversion
is slight and not clear-cut. Butyl amine alone gave a surprisingly
high yield. Ammonia, triethylamine, acetone and carbon dioxide as
additives had no substantial effect on yield.

t-
z
W
0
LL
W
4 2
LL
w
1.0
O.! 5
-
A o
1.5- d
z O//
A
- 92 3 '
@ STYRENE WITH I- BROMOBUTANE
A STYRENE WITH 2-BROMOBUTANE
0 STYRENE WITH 2-BR-2-ME-PROPANE
rn STYRENE WITH BROMOFORM
v)
LL
W
0 0
2 0.5- 0
0
0
k
z
W
0
a
W
-
a
a
W
2
3
0
a
0
I-
z
9
v)
a
w
>
z
0
0
I
I
0.1 0.2 0.3 0.4 0.5
MOLE FRACTION OF ADDITIVE
A.
A * A
A
Figure 3
* STYRENE WITH WATER
A STYRENE WITH BUTYL AMINE
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7
MOLE FRACTION OF ADDITIVE
Figure 4
b
A
A
/.
\
f !
\

ANALYTICAL RESULTS
Some evidence was collected on the chemical nature of some of the
products. The products deposited on the moving film and on the glass
electrode covers were always easily dissolved in common solvents like
acetone, chloroform and benzene, indicating low molecular weight and
that there was little cross-linking.
Infrared spectra were obtained of polymers made from styiene, benzene
and toluene by dissolving the deposit from the glass electrode cover
with chloroform and evaporating the solution on a salt plate. The
spectra were all similar and closely resembled conventional poly-
styrene. As Figure 5 shows, there were additional absorptions at
1050, 1220, 1720 and 3400 wave numbers indicating oxygenated products
of various kinds including hydroxyl or amino, ester and probably
ether. These spectra are very similar to those published by Jesch,
Bloor and Kronick (5).
The infrared evidence of oxygen-containing groups was obtained on a
styrene product which was prepared and transferred in a nitrogen
atmosphere. The only contact with air was during the time the spec-
trum had been run.
The sample was exposed to the atmosphere for 24
hours and it changed only slightly. As Figure 6 shows there was little
further increase of bands attributed to oxygenated groups. The only
noticeable change in the spectrum was in the relative peak heights at
700 and 760 wave numbers. The spectrum of a product made with l-bromo-
butane mixed with the styrene was almost identical to the unmodified
styrene product with a stronger absorption ascribed to ketone carbonyl
at I720 wave numbers.
The final evidence of chemical composition was elemental analysis.
The styrene product had considerably less carbon than styrene itself,
as seen in Table 3. It also contained a significant amount of nitrogen
and, as determined by difference, a large amount of oxygen. Heating
the polymer in air did not change the composition significantly. Sty-
rene polymer made in the presence of 25 to 30 weight-percent l-bromo-
butane had almost the same analysis plus a significant bromine content.
TABLE 3
Elemental Analysis of Corona Polymers
Monomer - C - H - N 0 (by difference)
Styrene, unheated product 72.23 6.06 3.2 - 17.71
Styrene. oven-heated 73.46 6.60 3.39 - 16.47
Styrene plus bromobutane
unheated product 73.74 6.95 3.59 6.00 9.64
oven-heated product 73.29 6.69 3.56 6.97 9.49
Styrene, calculated 92.26 7.74 - - -
Bromobutane, calculated 35.06 6.62 - 50.32 -

'603 ~---~ _-_- T_--r.-
Figure 5
1000 900 800 700 650
E - TOLUENE
F. BE NZENE
G -SlVRENE
i
\

DISCUSSION
289
The salient features of the information presented above might be
summarized as follows. In a mixture of organic vapor and nitrogen
subjected to high-frequency, high-voltage, electrodeless discharge,
the nitrogen, the organic compound and trace amounts of oxygen and
water are activated and combine chemically to yield products of higher
molecular weight than the starting materials. These proqucts condense
on any solid surface available and may even undergo further chemical
reaction within themselves and with more monomeric material which is
not electrically activated.
The chemical mechanism of this series of reactions must be complex,
possibly produced by simple ionization through collision with a
rapidly-moving electron followed by expulsion of two secondary
electrons:
N2 + e-+N2+ + 2e- or
+
RH + e---+RH + 2e-
I' These are the products ordinarily found in mass spectrometry and on
exposure to gamma or beta radiation. These activated cationic products
may activate vinyl polymerization or undergo secondary reactions,
u' yielding neutral free radicals or even anions, either of similar
structure or in the form of fragments and combination or rearrange-
9 ment products. In the present case there exists the further possibility
that reaction products become reactivated, since they are formed in
the presence of the high-voltage field, and undergo further reaction.
1 The simplest mechanism to consider would be polymerization of the
I vinyl monomers to long-chain products after initiation by some active
c
,, species. The fact that non-vinyl compounds gave good yields may be
' explained through a mechanism involving some fragmentation of every
1 monomer molecule and combination of the fragments to products of higher
molecular weight. The vinyl monomers could be reacting through this
P*
i
non-vinyl mechanism also.
The action of halogenated additives suggests that the yield of
reactive intermediates is increased by some of them and that these
intermediates increase the initiation of vinyl polymerization. There
is not enough data to .establish this firmly. It may also be consid-
ered that, by changing the dielectric constant of the gas mixture, the
additive can increase the efficiency of energy transfer from the elec-
tric field to the monomer.
The other additives had little or no effect, indicating that the
situation is not very similar to bulk or solution polymerizations by
5 free radical or ionic catalysis such as are now being actively studied
by several laboratories (6,7). Certainly we do not have a simple
vinyl polymerization. Otherwise some of the potential inhibitors
would have shown an effect.

230 . .
One of the unexpected results was t e fixation of nitrogen. This
can be seen in retrospect to be related to work with discharge-acti-
vated nitrogen which has been reported from time-to-time (8,9,10).
No complete mechanistic explanation will be presented. More experi-
ments are in process but it is not expected that a system which is
as potentially complex as this can soon be completely explained.
ACKNOWLEDGEMENTS
The able assistance of Messrs. Roger Kolsky and Walter Miner and
the helpful analytical interprelations of Mr. Elliot Baum are grate-
fully acknowledged.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
a.
9.
10.
A. Charlesby, “Atomic Radiation and Polymers, “ Pergamon Press,
Oxford, 1960, p.1.
R. M. Brick and J. R. Knox, Modern Packaging, January 1965,
pp. 123-128.
T. Williams, J. Oil and Colour Chemist’s Association 2, 936-951
(1965) .
Manmade Textiles, July, 1965, pp. 58-9.
K. Jesch, J. E. Bloor and P. L. Kronick, J. Polymer Sci.,
A-1, 4, 1487-97 (1966).
R. C. Potter, R. H. Bretton, D. J. Metz, J. Polymer Sci. A-1,
2295-2306 (1966).
K. Ueno, K. Hayashi, S . Okamura, J. Polymer Sci. BA, 363-8,
(1965).
L. B. Howard and G. E. Hilbert, J. Am. Chem. SOC. 60, 1918
(1938).
T. Hanafusa and N. N. Lichtin, J. Am. Chem. SOC. 82, 3798-9,
(1960) .
G. D. Steinman and H. A. Lillevik, Arch. Biochem. Biophys.
- 105, 303 (1964): CA 61, 4198 (1964).
I

? ~
MENBERSHIP
IN THE DIVISION OF FUEL CHEMISTRY
I The Fuel Chemistry Division of the American Chemical Society is an internationally
!recognized fol'um for scientists, engineers and technical economists concerned with
I
the conversion of fuels to energy, chemicals, or other forms of fuel. Its interests
center on the chemical problems, but definitely include the engineering and economic
aspects as wen.
Any chemist, chemical engineer, geologist, technical economist,,or other scientists
1 concerned with either the conventional fossil fuels, or the new high-energy fuels--
' whether he be in government, industry or independent professional organizations-would
benefit greatly from participation in the progress of the Fuel Chemistry
I
Division.
The Fuel Chemistry Division offers at least two annual programs of symposia and general
papers, extending over several days, usually at National Meetings of the American
! Chemical Society. These include the results of research, developnent, and analysis in
t the many fields relating to fuels wbich are so vital in today's energy-dependent
economy. Members of the Division have the opportunity to present papers of their own,
or participate in discussions with experts in their field. Most important, the Fuel
Chemistry Division provides a permanent record of all of this material in the form of
' complete preprints.
\\
I
Stding in September 1959, the biennisl Fuel Cell Symposia of the Division have been
the most important technical meetings for chemists and chemical engineers active in
,,this field. These symposia have all been published in book form. The recent land-
,mark symposium on Advanced Propellant Chemistry has been published in book form also.
J '
I Further, the Division is strengthening its coverage of areas of air and water pollu-
4 tion, gasification, and related areas.
In addition to receiving several volumes of preprints each year, as weU as regular
news of Division activities, benefits of membership include: (1) Reduced subscrip-
;" tion rates for '~uel" and "Combustion and Flame," (2) Reduced rates for volumes in
the "Advances in Chemistry Series" based on Division symposia, and (3) The receipt
card sent in acknowledgment of Division dues is good for $1.00 toward a ccmplete set
,',of abstracts of all papers presented at each of the National Meetings.
,I
To join the Fuel Chemistry Division as a regular member, one must also be or becme a
member of the American Chemical Society. Those not eligible for ACS membership
, because they are not practicing scientists, engineers or technical econdsts in areas
related to chemistry, can become Division Affiliates. They receive all benefits of'a
$)regular member except that they cannot vote, hold office or present other than invited
papers. Affiliate membership is of particular value to those in the informational. and
library sciences who must maintain awareness of the fuel area. Non-ACS scientists
'active in the fuel area and living outside of the United States are invited also to
\pecome Division Affiliates.
Membership Fn the Fuel Chemistry Division costs only $4.00 per year, or $pL.oO for
in addition to ACS membership. The cost for a Division Affiliate, with-
ACS, is $10.00 per year. For further information, write to:
Dr. Frank Rusinko, Jr.
Secretary-Treasurer
ACS Division of Fuel Chemistry
c/o Speer Carbon Ccunpany
St. ~ y s Pennsylvania ,
15857
Telephone: 814 - 834-2801

i
.;
Chemical Engineering Aspects of %?&mica1 Synthesis
in Electrical Discharges.
J.D. THORNTON.
Department of Chemical Engineering, The University,
Newcastle upon Tyne, England.
I 1.0.Introduction.
Although numerous investigations have been published dealing with the effects
of electrical discharges upon chemical reactions, most of the work is fragmentary
1
,I
'
r/.
in the sense that even now no clear picture has emerged of the relative importance
of discharge characteristics and reactor geometry upon the reaction yield. This
is largely because no systematic studies have been undertaken with a fixed
reaction System covering a reasonably wide range of discharge conditions and
reactor configurations. Furthermore most of the published work has been concerned
with laboratory-scale investigations so that, with the possible exception of ozone
synthesis, little or no attention has been paid to the chemical engineering problems
involved'in gas phase electrosynthesis. This is no doubt consequent upon the low
yields obtained in many laboratory studies and is unfortunate in that a proper
application of the engineering factors affecting the reactor efficiency could, in
many Cases,. lead to considerably improved reaction yields.
Whereas physical chemistry is usually concerned with the individual kinetic
/ steps contributing to a particular reaction scheme, chemical engineering is
concerned with the translation of the laboratory Teaction to a full scale contin-
! uous process wherein the desired reaction product is produced at the lowest capital
' and operating costs. In the case of gas discharge synthesis, these financial
< criteria imply that the reactor should have three highly desirable characteristics:
(a) The energy yield (Gms. product per Kw.Hr.1 should be high.
(b) The con\.ersion per pass should be high.
d
/
(c) The reactor should be selective for the desired product, i.e. side
reactions with their consequent wastage of material and energy should be
minimised.
In practical terms, it is not necessary to have a detailed step by step know-
,p'ledge of the reaction kinetics in order to design the reactor, so long as an
{ I
empirically determined rate equation is available which adequately describes the
influence of the operating variables upon the net rate of reaction. Clearly a more
informed approach to the design problem is possible when fundamental data are
' available but this is seldom the case at the design stage.
,d Assuming that the electrons in a discharge are able to excite the reactant
molecules, it is possible that many of the low product yields hithertoreported may
E be attributed to one or more of the following factors:
(a) The use of unsuitable reactor geometries in which a sizeable fraction
J' of the reactant by-passes the discharge zone.
II
'Y
\
?
\
,
(b) The use of non-optimum discharge conditions in which the average
activation cross-section for the desired reaction is low.
(c),The use of inappropriate residence times or residence time distributions
of the reactants and products.
\/
1
(d) The presence of parallel reactions which compete for the reactant and
the occurrence of rapid reverse and degradation (or consecutive) reactions which
destroy the primary product.
j
It is with these aspects.of gaseous electrosynthesis that the present paper
is concerned, since it is possible that many of these effects can be minimised by
', proper consideration at the reactor design Stage.
T
, 2.0 Reactor Classificatlon.
I It is useful to conslder initially the basic reactor geometries that can be
,, employed in a contlnuous-flow reaction system. Although no formal reactor classlf-
/ lcation has hlthert0 been proposed, the electrode configuration and direction of
I
3

292
flow of the reactant gas stream can be made the basis of a convenient system of /
classification. Thus the electrodes may consist of parallel plates, co-axial
cylinders or a pair of points whilst the gas flow may be parallel or at right
angles to the plane of the discharge. Figure 1 shows in diagrammatic form the
4 \.arious electrode anti flow arrangements on the basis of this classification.
Parallel flow in parallel plate and coaxial electrode reactors necessitates the use
of porous electrodes through which the gas phase flowsas shown in Figures 1.1 and
1.2. On the other hand, the commoner cross-flow arrangements shown in Figures 1.4,
1.5 and l.G usually employ impervious metal electrodes, there being :no necessity
for the reactants to flow through the electrodes themselves. The usual parallel/ 3
point anti cross-flow/point reactors are depicted in Figures 1.3 and 1.7 respectively.
The distinction between parallel-flow and cross-flow is not an academic one
by reason of the non-uniform nature of the electric field in manyof these reactor
arrangements. In most of the laboratory investigations reported in the literature,
the areas of the electrodes are small and little or no attempt has been made to
minimise c.{lgr effects by the use of suitable electrode profiles. The field
intensity is therefore a function of position in the inter-electrode gap. Furthermore
the iise of coaxial geometries where the respective electrode radii are very
different also gives rise to highly non-uniform fields in the vicinity of the
central electrode. It follows therefore that the or.erall activation effects
might well be different in the parallel/parallel and cross-flow/parallel systems
in Figures .1.1, 1.4 and 1.5. In the former case, molecules traversing a centreline
path will encounter electrons of different energies to molecules following a
peripheral path whilst, in a cross-flow reactor, all the molecules encounter
electrons with a wide spectrum of energies. In coaxial reactors, the position is
further complicated by virtue of the non-uniform field associated with the central
electrode. Thus in Figures 1.2 and 1.6 not only is there a transverse field
variation but there is also an axial variation as the molecules approach the
bottom and top planes of the outer,plectrode.
Such considerations are purely speculative at the present time because no
experimental data are available which would enable a direct comparison of reactant
conversions for parallel and cross-flow systems to be made for the same reactants
under similar conditions. Nevertheless these factors must be born in mind when
interpreting the influence of reactor geometry on a quantitative basis.
Coaxial electrode systems in which the reactant flows through the annular
space between the two electrodes have an advantage in that all the reactlnt
molecules must pass through the discharge. There is therefore no by-passing and
subject to suitable electron energies, there will be a high level of activation
in the gas phase. Such ccnditions are difficult to achieve in parallel plate
arrangements unless the reactant is introduced at the centre of one of the electrodes
as in Figure 1.4. Much work has been reported with point electrode Systems
such as those illustrated in Figures 1.3 and 1.7 and here it is difficult to see
how by-passing of the discharge can be avoided. Furthermore the flow pattern
within the reactor is also dependent upon the thr.?ulence level so that changes
in the gas flowrate will almost certainly be accompanied by corresponding changes
in the fraction of the gas stream by-passing the discharge zone.
All the reactor systems discussed so far can be described as homogeneous in
the sense that only a single, gaseous, phase is present in the reactor. In many
instances, however, it is advantageous to remove one or more products of reaction
as rapidly as possible' in order to minimise subsequent reactions which would
adversely affect the primary rehction yield. A convenient method of achieving
this involves the introduction of a second.phase into the discharge zone in the
form of a liquid absorbent (6) or fluidised solid adsorbent (\&I. In principle
a second phase can be introduced between any of the electrode arrangements shown
in Figure 1. This modification gives rise to a second series of reactor-systems
which will be referred to subsequently as heterogeneous reactors.
3.0 Nature of the Electrical Discharge.
For optimum reactor performance it is necessary to ensure that the reactants
\
I
i

ELECTRODE ARRANGEMENT
PLRALLEL CO-AXIAL I POINT
I- Fig-1 ~
Ihsir:
rcaclor p?ornrrtrics.
-
i t

294
have suitable residence times in that part of the discharge zone where activation
is most favoured. In this connection it will be recalled that the relative rate '
of ammonia synthesis from its elements has been shown to be greatest in the region
of the cathode potential drop (3,) whilst in the case of hydrazine synthesis from
ammonia, the significant fraction of the discharge is the positive column region
(k). In theory, it is desirable to i
be able to predict the discharge conditions
necessary for optimum conversion in any particular reaction; in practice however,
this is not yet possible. This is-because in the past it has been customary to (
describe discharges in general terms such as glow, arc or silent digcharges without
regard to the local or point properties existing in different parts of the discharge;
zone. Since there is a gradual transition between our regime and the next, qualit-
ative descriptions of this type are not of great value in interpreting reaction
rate data. Any fundamental correlation of rate data with discharge characteristics
would involve not only a knowledge of the concentrations and energy distributions
of the various particle -species present in the discharge space)including electrons
but also an understanding of the way in which the probability of excitation varied 'k
with electron energy . A typical example of the way in which the degree of
excitation is dependent,upon electron energy is shown in Figure 3 which depicts .
total cross-section curves for argon and neon ( 5). The cross-section for an mn -)
type transition by electron collision is zero when the electron energy is less than
the energy required to raise a valency electron of an atom from the mth to the \
nth level,-but increases with electron energy to a maximum after,which any further
increase in electron energy is accompanied by a decrease in the value of the
total cross-section. There is therefore an optimum value of the electron energy
corresponding to the maximum cross section for the process under consideration. 1
It is the lack of data of this type together with a knowledge of the reactive
species present in the discharge that makes it difficult to correlate reaction \
rates with discharge parameters for systems of engineering interest. I
Nevertheless despite the absence of fundamental information relating to the
\
.activation characteristics of discharges, one or two general observation can be
made with regard to the nature'of the discharges commonly employed. All the
reactor geometries discussed so far, and illustrated diagrammatically in Figure 1,
employ the same type of discharge in the sense that both electrodes are located
within the reactor and are therefore in contact with the reactant gas. This class
of discharge in which electrons pass freely between the two electrodes is represeh
schematically in Figure 2a. An alternative type of discharge which has been used
I
frequently is the so called barrier discharge shown diagrammatically in Figure 2C.
Here one of the electrodes is separated from the gas phase by means of a dielectric)
barrier such as quartz so that the reactor may be looked upon as a capacitative '
and resistive load in series. In practice a dielectric barrier may be employed \
in conjunction with any of the reactor geometries shown in Figure 1. This
arrangement has the advantage of providing a more uniform type of discharge which
is current limited so that the build up of spark or arc type discharges is avoided.\
If this principle is carried to its logical conclusion, both electrodes may be
isolated from the reactant gas and molecular excitation set up by applying a high
frequency potential.to the external electrodes or by induction from an external \
conductor carrying a high frequency current.(Figure 2d.) This type of "electrodeless:
reactor has not been mentioned in the classification discussed above but is Of
interest since in the absence of metal electrodes. in the gas phase, the degree Of
dissociation of the gas and therefore the concentration of atoms would be expected
to be high (LO). The high concentration of atomic species in such discharges is
evidenced by the afterglow phenomena frequently encountered in electrodeless
systems.
One type of electrodeless sytem which might well prove of interest in eleCtrO-;
synthesis is' the microwave discharge. Here the degree of activation can be highcq)
and it has been reported that the atomic yields in hydrogen, nitrogen and oxygen 1
are approximately ten times greater than the yields of atoms and radicals in low
frequency and direct current discharges at the same field strength and pressures.
From the bngifieering point of view, microwave systems have the added advantage
1
,I
-I
t
'
i'
c
S

295
'' that they can be sustained at higher pressures but on the other hand the detailed
,t economics still require'evaluation.
The crossed discharge arrangement shown in Figure 2b comprises a low frequency
discharge at right angles to one of high frequency. The discharge zone itself is
said to be characterised by a low energy density and to result in unexpectedly
high activation of the reactants and relatively high yields of the reaction products
(3 ). These claims ha$-e not however been confirmed by recent work in these
j
laboratories in which methane was subjected to a crossed discharge at 31 mm.Hg
pressure. The reactor consisted of a 15 mm. internal diameter glass: tube fitted
i/ with two pairs of 3 mm.diameter stainless steel electrodes arranged at right angles.
The percentage decomposition of methane was observed using a 0.91 Mc/s high
frequency discharge, a 50 c/s low frequency discharge and combinations of high
1. and low frequency discharges simultaneously. Figure 4 shows some typical data
plotted in the form'$ methane decomposed versus $ H.F. power for a series of runs
at constant total power. The methane flowrate was constant throughout at 93 cc/
'
minute,at N.T.P. These results confirm the general trend whereby an increase in
disch,arge power is accompanied by an increase in the percentage decomposition of
the methane but on the other hand do not substantiate the claim made for the higher
activating effect of the crossed discharge. Had this been the case, the curves
shown in Figure 4 would have exhibited maxima whereas, in fact, the observed
methane decomposition at constant total power increased steadily with increasing
percentage pf H.F. power. In this particular case, there is therefore no advantage
in the use of crossed discharges and the characteristics claimed by Cotton still
lemain to be confirmed . -
4.0 Product Selectivity and Residence Time Considerations.
In the simple plug flow reactor, the conversion (x) obtained is related to the
residence time (f) by the expression:
. .
"where U, and r are the molal volume of the reacting mixture and the reaction rate
respectively. If the reaction is one which can only take place under the influence
i of an electrical discharge, then the residence time refers to the time spent by
1 the reactants in the discharge zone itself. If however subsequent reactions involv-
,;ng the primary reaction products are possible outside the discharge zone, these
I
$1. will usually proceed at a different rate and must be taken into account separately
in analysing the overall performance of the reactor system.
\ If the reaction rate (r) could be written in terms of the concentractions of
the reactants and a rate constant (k) which in turn was dependent upon the discharge
characteristics, then in principle it would be possible,to integrate the above
'I expression for any set of discharge conditions and compute the residence time for
! any desired con--ersion. In practice however this is seldom possible because the
functional relationship between the reaction rate and the discharge characteristics
'~
Dishcarge reactions are frequently complicated by reason of the interactions
) between the atomic, molecular, ionic and free radical species present. Moreover
because of the wide range of species present, reactions are rarely simple in the
sense that only one primary reaction product is produced. Competing parallel,
consecutive and degradation reactions are often present, the net result of which
is to produce a wide spectrum of reaction products. One of the key problems at
the present time therefore is how to design reactors in which unwanted side
reactions are reduced to a minimum. There is no clear-cut anawer to this p'roblem
as yet, but an examination of some of the more elementary types of reactions such
as those listed in Table I indicates a promising method,of approach.

0
E
<
h(
E
l!
d K
c 0
.-
Y
0
t
u
t
loo Po 300 100
Elrctron Energy, V
Fig. 3. Typical Cross-Scction versus Electron Biergy Curves.
High Frequency Power z.
Fie.4. Effect of a Crossed Discharge on % Metha*$ l’)nro-ipnnition.

\
,
,
~ ~~ -
Consecutlve Reactlons A + B A C
A+C-D
Polymerisation Reactlons A + A h B
A + B--.C
A + C-D
Parallel Reactlons
TABLE I
297
Examples of Elementary Reactions.
A + Bgg
Reversible Reactions A + B S C + D
I
Degradative Reactions A + B 4 C d D - E
In all the react,ions listed in Table I it is assumed that C is the
preferred product. Intermediate steps involving excited species are omitted so
that the equations merely represented the overall stoichiometry of the various
situations. If therefore product C is being produced from reactants A and B
and a second (consecutive) reaction is present which destroys C , then the rate
of the second reaction can be reduced by maintaining the concentration of C at
a low value.
Since the primary reaction responsible for the formation of C
should, on economic grounds, proceed as rapidly as possible, the only way of
maintaining a low product concentration is to remove C from the discharge zone
as rapidly as it is formed. The methods by which this can be achieved will'be
discussed further below.
The same principles apply to the other reactions in Table I. Thus in the
case. of polymerisation reactions, the molecular weight of the product can be
controlled by selective removal of certainconstituents from the discharge. If
product B is preferred, then further polymerisation can be minimised by rapid
removal of B. If on the other hand, the higher molecular weight product C is
required, B is allowed to remain in the discharge and C is selectively removed.
It frequently happens that the product yield is limited by reverse and parallel
reactions. In the latter case, C can be produced in preference to D by
selective removal of C. In the former instance, any yield limitations imposed
by equilibrium. considerations can be overcome by rapid removal of C so that the
equilibrium is displaced almost entirely in favour of products.
Similar reasoning shows that in the case of degradative reactions, product
C can be produced in preference to D or E provided that C is removed rapidly
enough from the activating influence of the discharge.
A number of methods are available for the rapid removal of a particular
component from the discharge zone, namely:
(1) Quenching the reaction at low temperatures and freezing out the product (x)
(2) Absorbing the desired product selectively in an inert liquid absorbant (6)
(13 1
(3) Adsorbing the desired product in a fluidised bed of solid adsorbent (I&).
(4)'By using a low residence time in the discharge zone. This may be achieved
either by high gas flowrates or by using a pulsed discharge technique. These
are all devices which reduce both the produce concentration and -residence time in
the gas phase and so minimise product losses through subsequent reactions.
Methods (1) - (3) are means of controlling selectively the residence time and
concentrations of certain specific products in the discharge and a choice can only
be made between them in the light of the physical properties of the various
constituents of the gas phase. The fourth method, which does not discriminate
between the concentrations or residence times of reactants and products, can only
be evaluated when the relative rates of the competing reactions are known. Of the
various techniques available, simultaneous reaction and absorption in a heterogeneous
reactor (method 2) opens up interesting possibilities for producing economic
I

298
yields of many chemical products where hitherto the overall yield has been
restricted by decomposition reactions. The theoretical analysis of such systems
is howe!.er extremely complex and even if the kinetics of the process were fully
understood, it is doubtful if the absorption rate into the liquid phase could be \
predicted with any degree of certainty because of complications arising from
dielectrophoretic stirring of the liquid. If the liquid phase is polarizable
the electrical forces in a non-homogeneous field will cause polar molecules to
move towards regions of maximum non-homogeneity. Any turbulence thereby set up I
within the liquid phase will enhance the rate of gas absorption but at the same
time render the prediction of. mass transfer coefficients extremel-J uncertain.
From a practical point of view, such effects are desirable insofar as it is
important that the rate of absorption of the product should be high compared with
its rate of production by synthesis.
Some illustrations of the factors -d'iscussed are of interest at this stage and
Figs.5 and 6 show how the molecular weight of the product is influenced by
residence time in the.case of consecutive type ractions involving methane as the a
reactant. Figure 5 shows how the higher molecular weight product, in this case
propane, is favoured by longer residence times in the discharge (11 ). These data
were obtained at 80 mm Hg pressure in a cross-flow/coaxial reactor with a 1.60 mm. \
thick quartz barrier adjacent to the inside surface of the outer electrode. The
relevant electrode diameters were\-l mm and 3 mm respectively and the power \
density was0 .O&w/cc throughout. Figure 6 shows the effect of longer residence
times on the higher molecular weight fraction from methane at a pressure of
10 mm Hg and again the molecular weight of the product is strongly dependent on
residence time (7 ).
In the discharge synthesis of hydrazine from ammonia, it has been suggested
( l.+ )( I ) that the hydrazine can be subsequently destroyed by further electron \
excitation or by reaction with atomic hydrogen produced in the primary electron I
bombardment of ammonia. If therefore th' concentration and residence time of
ttie hydrazine in the discharge be reduceLydegradative reactions should be
and the yield increased correspondingly. Tha.t this is the case can be seen from
Figures 7 and 8i The arrangement of the cross-flow/coaxial reactor employed in these
runs is shown in Figure 9. The reactor tube was made from 1.60 m.thick quartz
tubing which constituted the dielectric barrier and was 28.6 mm.0.D. The inner
electrode consisted of a perforated stainless steel drum 18 mm O.D. which was
mounted on a coaxial shaft so that it could be rotated at high speed. A liquid
absorbent, in this case ethylene glycol, was fed to the drum so that, as the latter.
rotated, the glycol was forced.through the drum perforations into the discharge
annulus.in the form of a fine spray. The data shown in Fi,gure 7 were obtained at
an average power density of 0.044 Kw/cc and illustrate the fact that not only can
the energy yield of h7drazine be increased'by reducing the gas residence time
but that a further substantial increase in yield can be obtained if the hydrazine
is rapidly removed from the discharge by selective absorption (13). at
The data shown in Figure 8 in which the energy yield is plotted versus the
discharge duration for a constant gas flowrate were obtained in a cross-flow
parallel plate reactpr using a pair of electrodes each measuring 12.5 x 6.3 mm.
and located llmm. apart. 1n.thes.e runs the average power density was 0.01 Kw/cc
using a pulsed D.C.discharge, the pulse duration ranging from 12 to 130 micro-
seconds. Under these conditions, the effective,residence time of the reactant and
products was no longer dictated by the gas flowrate but by the pulse duration.
This technique is therefore a device for securing very low residence times without
the inconvenience of high gas velocities. Again the data are in agreement with
previous results in that the lower residence times favour higher energy yields
because of the smaller loss of hydrazine through reverse and consecutive reactions.
It is interesting to note that an alternative technique whereby atomic hydrogen is
converted to less reactive molecular hydrogen through the agency of a platinum
catalyst has'also shown increased yields of hydrazine ( k).
.It has been assumed 60 far that the reactant flows as a coherent plug
throu& the discharge. ?his concept of plug-flow is only an approximation at
I
1
, I
I
\ '
\c
<
4
i
i
t
c

05 1.0 15
5
Residence Time 7x10 hr.
Fig.5. The effect of residence time on propane-ethane yield
from methane.
Hipher Mobcular Welght FrCCtia?
1 2 3 4
Rnldrncr flmr rlO' hr
-
Pressurr lOmm -
rig.6. The cffr~t
of rr,sidcnr.rA time on the yield of higher
molrvular wc.ight fraction from methane.

- ;i
Y .
15. Pressure lOmm
5
> 10- 1
--
4 U _,
a
2-
e
c 5.
N
0
k
I ,.
-.

J
I
i'
i P
L iqud
i
Gas

302
the best and in many practical situations considerable divergences are observed I
between the residen

Acknowledgements .
303
The author is indebted to a number of his colleagues and to Dr. C.A.Walley
of the Department of Electrical Engineerlng for manv helpful discussions.
Refcrcnrcs.
(1) Andcrsen W . H . , Zwollnski B.J., Parlin R.B., 1nd.Eng.Cheni. ,1959.%.(4).527.
Breirer A.L. , W~sthai cr J.W. , J .Phvs .Chem. , 1929.2.883 (Cf. also J .Phys .Chem.
1930. '4. 153)
Cotton W. J. , Trans .Elect rocheni.Soc. , 1947. 2, 407, 419.
Devins .J .C. , Burton M. , J .Anier.Chem.Soc. , 1954., 3. 2618.
Ilowatson A.M., "Introduction to Gas Discharges" (Pergamon. PreSS,1965).
Imperial Chemical .Industries, U.K.Patents 948,772; 958,776; 958,777; 958,778;
9GG,40G (all 1964).
Kraaij\.cld Von H.J., Waterman J.I., Brennstoff-Chemic?, 1961.42.(12),369.
Llewellyn-Jones F., "Ionisation an? Breakdpwn in Gases", (Methuen & Co.
Lrd. 1966).
AlcCarthy R.L., J.Chem.Phys., 1954, 22, 13FO.
hliya-zaki, Takahashi, Nippon Ragaku Zasshi, 1958. 3. 553.
Sergio R. Unpublished Work, University of 'Newcastle upon Tyne. England.
Thornton J.D., Chem.Proccssing, 1966. (2), SG.
Thornton J.D., Charlton W.D., Spedding P.L.., "Hydrazine,Synthesis in a
Silent Electrical Discharge" (to be published).

GENERATION AND MEASUREMENT OF AUDIO FREQUENCY POWER
FOR CHEMICAL-ELECTRICAL DISCHARGE PROCESSES \
James C. Fraser '.
Research and Development Center
General Electric Company
Schenectady.. New York
In developing equipment for use in high voltage, high frequency
chemical-electrical processing, we have built a number of different
types of "corona" power supplies. At the start, from the standpoint
of the electrical equipment, let us use "corona" in its broadest sense.
We will define a "corona" discharge as an electric discharge produced
by capacitively exciting a gaseous medium lying between two spaced
electrodes, at least one of which is insulated from the gaseous medium
by a dielectric barrier. A corona discharge may be maintained over
wide ranges of pressure and frequency, although atmospheric pressure
and frequencies in the audio range substantially above power trans-
mission values are typically employed.
Since the power which can be dissipated in a corona reactor or
cell is a function of the supply frequency, we have settledarbitrarily
on the audio range (3,000 to 10,000 cycles per second) as a desirable
compromise. It is within the l i m i t s of rotating machinery and solid
state components, presents no undue corona reactor heat dissipation
problems, and provides a happy compromise between operating voltages,
reactor size, and economy of operation.
The ratings for the basic electrical components are dependent
upon the characteristics of the corona reactor and the impedance load
which it represents to the power supply. Corona reactor values which
affect these ratings are:
a. The corona power required
b. The electrode corona power density and, thereby, the
electrode cooling or heat dissipation capability
C. The gaseous atmosphere present
d. The gaseous gap spacing
e. The barrier material and dielectric constant
From these values we can determine:
a.
b.
c.
Peak voltage required to initiate corona
Barrier thickness required to withstand the total voltage
across the barrier in the event of an arc across the gap
Barrier, gaseous gap, and total reactor capacitance
d.
e.
f.
Maximum operating voltage
Capacitive charging, or displacement, current drawn by
the reactor
Corona resistive current .
g.
h.
Resultant load current
Power factor of the corona reactor load
I .I
1
?
\
1
\
'\

1
f
/
I 7
\
i/
'1:s
The objectives which,we attempted to meet in the design of our
equipment were :
a. Suitability for use with a wide variety of corona reactors
b. Operation over a wide range of breakdown voltages
C. Flexibility over a wide range of operating conditions
d.
e.
Maximum electrical efficiency
Trouble-free operation and minimum maintenance :over long
periods of ,operation
Among the types of equipment built were the inductor-alternator
with step-up transformer and parallel-resonant tuned circuit, the
SCR inverter with step-up transformer, and the mobile vacuum tube
high power-amplifier.
Motor-Alternator Set
A typical power supply based on rotating equipment consists of
the following major components: an inductor-alternator, by means of
which 60 cycle power is converted to 10,000 cycles, at 500 volts.
This output is fed to the primary of a high voltage, high frequency
transformer rated for 25 or 50 KV at 10,000 cycles. The output of
this transformer is fed, in turn, to a parallel resonant tuned circuit,
the discharge reactor, and the high volthge instrumentation.
The KVA rating of the hl-G Set (inductor-alternator) is determined
by the product of the corona operating voltage and the resultant load
current (or the corona power divided by the power factor). Since the
corona power generated in a gaseous gap in series with a dielectric
is directly proportional to frequency, and the upper l i m i t for frequency
for commercially available rotating equipment is about 10,000 cycles
per second, this has generally been considered the optimum frequency.
Also, the use of lower frequencies would require higher voltages or
increased corona reactor capacitance for any given power level. The
output voltage of the alternator can be any value consistent with usual
generator practice and with the transformer primary voltage.
The KVA rating of the transformer will usually be the same as
that of the inductor-alternator. Its secondary voltage will be de-
termined by the corona reactor requirements for voltage breakdown of
the gaseous gap. Secondary or load current will be the vector sum
of charging, or displacement, current and resistive, or corona, current.
Primary voltage must be consistent with alternator output voltage.
If experimental flexibility is required, the transformer primary should
be arranged for series-parallel connection.
f The corona reactor represents a capacitive load on the major items
1 of the power supply equipment, up to the time of corona initiation.
5 Since this generally results in a poor power factor, there are two
, possible means of solution:
Y
a. The motor-alternator set must be built with sufficient KVA
\ capacity to meet the power factor and corona power requirements.

. The transformer secondary and the corona reactor can be tuned
with a parallel resonant circuit consisting of a choke and
high voltage, vacuum capacitors. This provides power-factor
correction so that the high voltage transformer secondary sees
mainly the resistive load represented by the corona power.
The parallel-resonant circ.uit must be tuned to give maximum trans- 1
former output voltage for some arbitrarily selected value of alternator
field current. With a 0.16 henry choke, for example, the capacitance <
required for tuning at 10,000 cycles can be found from the expression 1
1 1
f = , or C = = 1580 picofarads
2 f G 462 €%
However, this value is reduced by the capacitance of the discharge react
itself, by that of any instrumentation voltage dividers, and by stray “t;
capacitances, so that the actual tuning capacitance used is somewhat
lower and must be determined experimentally. 7
The decision as to whether the motor-alternator should be built
with increased KVA capacity or whether the secondary circuit should be
tuned depends on the experimental I versatility required, and on economic
considerations. In general, the dollars per KVA for high frequency
motor-alternator sets are considerably higher than for tuning circuit
components.
An alternator field-control chassis supplies and regulates the current
for the alternator field windings, and has a feedback system designea
to meet varying conditions of high voltage output. Alternator output
voltage and, thus, transformer high voltage are raised and lowered by
changing alternator field current.
A magnetic amplifier stabilizer
circuit maintains constant high voltage output. In the event of a de-
crease in high voltage, it calls for an increase in alternator field
current, with a corresponding output voltage increase.
I
\
1
‘
i
Control devices and instrumentation for a ’
corona power supply based
on rotating equipment includes at least the following: 1
a.
b.
C.
d.
e.
Provision for varyi .ng the transformer output high voltage from
zero to maximum by varying alternator field current, and a
means of measuring the field current.
Provision for measuring transformer primary voltage, current,
and power.
Provision for measuring the peak voltage applied to the corona
reactor.
Provision for measuring the power dissipated in the corona
reactor.
Some form of protective current overload relay, for removing
high voltage from the corona reactor in the event of a short
circuit in the reactor, or a low voltage,high current arc
5

i I.
r
>-
.i
between the electrodes. This will prevent exceeding the current
1' a t i n g o f t he t r a n s f o rme r s e c o n da r y .
Since the use of this equipment involves working with high voltage.
the control circuit includes provision for test area safety interlocks
and flasher-warning lights.
I nver t e r
j. The solid state corona generator consists of a 3.000 cycle inverter
I - utilizing silicon controlled rectifiers feeding a step-up transformer.
A rectifier is a. device or circuit for changing alternating current to
' direct-current. The inverter may be thought of as a rectifier operating
ill reverse, changing d-c to a-c. A combination of the two achieves a
q' change from 60 cycle a-c to approximately 3,000 cycle a-c.
.-'
Rectifier circuits occur in several configurations such as halfwave,
full wave, bridge. etc. Inverter circuits may be gr0u.p ed in a
/similar manner. The key feature of an SCR (silicon controlled rectifier)
as used in an inverter circuit is that a small current from its "gate"
4 element to the cathade can fire or trigger the SCR so that it changes
8 from being an open circuit into being a rectifier.
The inverter operates from a single phase, 220 volt, 60 cycle
I source, and by means of a full-wave bridge rectifier, rectifies it to
' 187 volts d-c which in turn is changed to 150 volts, 3,000 cycles,
1 approximate sine wave through SCR's. Variac control of the 220 volt
input provides a variable supply output.
The inverter output is fed to a step-up transformer delivering up
4 to 12,000 volts rms. This transformer is a dry type, potted with RTV,
with integral high voltage terminals. The particular transformer used
was built as a compromise between low cost, size, and electrical
capability. Rated at 3 KVA, the'transformer was only 6" x 6" x 7"
1 high overall.
I What is noteworthy with this equipment is that the inverter,
transformer, and corona reactor are not independently operating com-
' ponents. The concentric cylinder- corona reactor was so designed that
j its capacitance, refle'cted to the primary of the transformec formed
a part of the inverter LC circuit, as did the inductance of the high
' voltage transformer. The reactor gaseous gap conductance forms a
/part of the inverter load circuit.
1
Since this is an untuned high voltage circuit, the inverter and
) the transformer were designed to carry the power factor loss current.
r'
Y
\
Advantages of .the inverter supply over the motor-alternator is the
absence of moving parts, ease of mounting and installation, and the
"module" or building block concept whereby units can be stacked to in-
crease voltage and current capability.

Vaccum-Tube Amplifier
The mobile corona generating equipment was designed for operation
ove mr a wide range of frequency and voltage. Corona power is generated
by a vacuum tube, high-power amplifier driven by a stable low-power,
audio-frequency sine-wave oscillator. Power is taken from the secondary 1
winding of a high-voltage step-up transformer, with one end at ground
potential.
I $
5
The corona output voltage may be varied from zero to 50,00O:volts, 1
depending on the impedance of the applied load, and on the settings \
of the frequency and amplitude controls located on the master sine-wave
oscillator panel. The frequency of the output voltage is determined
by the master oscillator and is independent of load impedance or output !
voltage level. The frequency may be varied instantaneously between ,
the rated limits of 3 to 10 kilocycles per second.
The a-c power from the 60 cps line is converted into the high
voltage variable frequency output as follows:
A master oscillator feeds a driver amplifier. This is turn drives
a tetrode power amplifier stage and the output step-up transformer.
A d-c plate supply source, a screen grid voltage supply, a control grid
bias supply, and an a-c filament supply feed power at appropriate
voltage levels to the tetrode amplifier stage. The power line control )
circuitry incorporates on-off switching, circuit breakers, and starting
contactors for the power amplifier supplies.
Power amplification is produced by four tetrode vacuum tubes con-
nected in pushpull-parallel as a Class AB2 amplifier. Control of the *
master audio oscillator signal amplitude is provided by a potentiometer
on the master control panel. The output voltage of the equipment may
be reduced to zero even when full voltage is applied to the plates of \
the power tetrodes.
Power delivered by the tetrodes to the output transformer, and
thence to the corona reactor load, is controlled by the oscillator and )
the variable autotransformer in the tetrode plate power supply. The
autotransformer control keeps the total power consumed by the equipment
during an experiment as low as possible, to minimize heat developed
in the cabinet and to extend tube life. -.
The corona cell impedance, which is mainly capacitive before corona '
breakdown, acquires a resistive component after breakdown which depends (
on the rate of production of ionized and electrically conducting I
chemical entities. The a-c voltage at the plates of the power tetrodes 'i
will lag the plate current by a large angle before power is consumed, 1
and the plate current will be collected at a high rather than a low
value of plate voltage. Thus, the power dissipated by the tetrodes
will be high under no-load conditions.
The high voltage transformer has an effective primary to secondary '
step-up ratio of 11 to 1. The maximum a-c current that can be drawn
from the secondary winding of the transformer is 60 milliamperes rms.
$ 1
\
7
\
I
P
1
8
'u;
\
i
\
i

\ /
’\
i Reactor
I
300
The corona reactor or cell represents a capacitive load to the power
supply equipment up to the time of initiation of the electrical discharge.
The total reactor capacitance consists of the series combination of di-
electric barrier and gaseous gap capacitance. The voltages appearing
across the barrier and the gap divide inversely as their, capacitance.
’ Before corona starts, the reactor current consists of the charging,
or displacement, current which is proportional to the applied voltage,
/ frequency, and cell capacitance. After the start of corona, the cell
I acts as an impedance load consisting of a conductance across the gaseous
gap, in addition to the gap capacitance. The total current drawn by
the reactor then is the vector sum of charging plus resistive currents.
If i
the corona reactor power supply has a tuned circuit in the transformer
high voltage secondary, for power factor correction, the corona
reactor capacitance functions as part of the tuning capacitance, and
,’ the reactor charging current is a part of the circulating current in
the parallel resonant circuit. The total load current through the transformer
secondary winding then consists chiefly of corona current plus.
whatever capacitive current might be present as a result of tuning unbalance.
If the power supply is untuned, the total power available for
chemical processing is the product of supply volt-amperes times the
power factor of the load, and in this case transformer secondary wind-
ing must carry the full reactor impedance current.
The dielectric barrier in the corona reactor serves as a built-in
current-limiting device, or ballast in the event of a short circuit or
arc between the cell electrodes, so that the stored energy is dissipated
in the barrier at breakdown.
The maximum power density at which a corona reactor can be operated
without puncturing the barrier is a function of the dielectric strength
of the barrier material, the total thickness of the barrier, the ambient
temperature in which it is operated, and the gap spacing between
barriers. The dielectric constant of the barrier affects both the
voltage gradient across the barrier, and the corona power which can be
dissipated in the reactor. The dissipation factor is a measure of the
electric losses which produce heating of the barrier.
The voltage at which breakdown will occur between a set of electrodes
in a corona reactor depends upon the gaseous atmosphere present, the
density (pressure, temperature) of the gas between the electrodes, and
the gap length.
The corona power dissipated in a gaseous gap in series with one
or more dielectric barriers can be calculated from the expression
P = Corona power in watts

310
f = Power supply frequency in cycles/second
cb = Dielectric barrier capacitance
Cgas = Capacitance of corona cell gas gap
Ct = Total corona cell capacitance
I
Vgas = Peak voltage across the gaseous gap at the time of corona
initiation
Vmax = Total operating voltage applied to the corona cell
This can be re-written as:
Lgas
Vt = - = voltage across the corona cell at the instant
.Ct 'gas of corona initiation
An- item of concern to all, in evaluating an experiment, or in
scaling up for pilot plant operation, is the electrical efficiency of
a process. Care must be taken not to confuse equipment power factor
with what might be called "electrical efficiency of conversion",
both may contribute to the same end result. Power factor relates though\ to
ability to convert equipment input volt-amperes (not necessarily in
phase) to available chemical-result-producing "watts". This is a
function of electrical equipment and electrical circuit design.
i
"Electrical efficiency of conversion" may be considered as the
effectiveness in utilizing all or most of the available wattage in the
corona reactor. This is a function of reactor design, with particular
emphasis on relative barrier and gap capacitance and gap spacing. As (
a matter of practical design, the cell geometry and gap spacing present ~
limits to the power which can be dissipated in any given cell.
Power Measurement
The power consumed in the discharge cell is determined by means (
I
of the parallelogram-oscilloscope technique which shows the relation-
ship between the voltage on the cell electrodes at any instant, and
the charge flow in the circuit up to that instant. Before corona
initiation, the corona null value is seen on the oscilloscope as a
closed figure, or slant line. With the initiation of the electrical
discharge, the figure opens up into a parallelogram, and the discharge
power is represented by the area of the parallelogram.
The bridge circuit for power measurement consists of two branches 1
between the high voltage terminal of the reactor and ground, called
"Voltage" and "Charge". The "Voltage" branch consists of a capacitance
i
\ !
I
t
I
h
\
\

I
1
311
voltage divider giving an approximate lo4 to 1 voltage division. The
"Charge" branch consists of the corona reactor in series with some
value of Capacitance such that a voltage division of lo4 or lo3 to 1
will be obtained across the divider. The outputs from the low voltage
ends of these dividers are fed to the X and Y axes of the oscilloscope.
A variable potentiometer across the lower end of the "Charge" divider
provides phase shift control to bring the "Voltage" and "Charge" branch
outputs into phase for a null reading on the scope.
It should be noted that two signals, of equal amplitude, in phase,
fed to the X and Y axes of a scope, give a 45-degree line on the scope.
Any variation of relative amplitude shifts the angle of the line.
The area of the parallelogram represents the total power dissipated
in the cell when the parallelogram bridge circuit and the oscilloscope
are calibrated in the following manner:
The X-axis deflection is calibrated with a peak reading voltmeter
/so that reactor voltage (peak-to-peak) is presented as volts per cm
of scope deflection.
/
The Y-axis deflection is calibrated with the known value of
capacitance in the grounded end of the "Charge" branch so that the
charge flowing through the reactor is presented as coulombs per cm
/' of scope deflect ion.
/
The X - Y product, then, is:
-
) (vo:~s,)(cou:;bs 1
volts - coulombs, watt-seconds
cm2 or
cm2
/ . Since this is the energy per cm2 in one cycle, the power per cm2
of parallelogram area is obtained by multiplying the energy per cycle
2 by the frequency in reciprocal seconds. The parallelogram area can be
, determined by planimeter measurement of a Polaroid picture, or by
'J direct measurement.
\
With a dual beam oscilloscope it is also possible to view the
a"/ voltage and charge waveforms simultaneously.
\

FUFARCI! INSTITUTE OF TFYDLC UNIVERSTTY
b150 KNRY P.VE.
V'TIJDFLPPJA, PA.
INTRODIJCTICN
uses,
Dlasma penerators, in general, have been found suitable for a variety of
Thev penerally Drovide an electric arc which is condensed or constricted
into a smaller circular cross-section than would ordinarily exist in an onen arc-tyDe
device. This constriction penerates a very hinh temperature (8,000-20,000°K) so
that a sunerheated-Dlasna workinp fluid can he ejected throuph the nozzle and the
comoosition of the Dlasma determine the use to which the plasma nenerator is put.
Plasma Fenerators have been used for cuttinn, welding, metal spravine and chemical
orocessinr. For chemical processinp, plasma penerators have provided the possibility
of the production of new aaloys and compounds and the processing of less commonly
used materials, as well as the preoaration of certain common chemicals.
nLAS#A JET EQUTPMrNT
Two types of plasma generators are possible: the nontransferred and the
transferred arc. A nontransferred arc consists of a cathode and a hollow anode
where the arc is struck between the electrodes and the flame emerges through the
orifice in the anode, In the transferred arc, the cathode is placed some distance
away from the anode and an arc is passed between the electrodes. The nontransferred
arc is the nost popular in the'chemical studies made to date.
A plasma jet used in chemical synthesis can have varied desips to meet 1
special requirements, such as
oath at a particular point.
the introduction of a reactant material into the flame
Consumable cathodes have been used in experiments in which carbon was one of \
the reactants, Carbon, vaporized from a graphite cathode, was used in the synthesis
of cyanogen and hydrogen cyanide. powdered carbon introduced in a gas stream or as a
constituent of a Kas has also been used as the source of carbon in a plasma flame.
Electrodes of 2%-thoriated tungsten are the most frequently used water-cooled
nonconsumable electrodes, Hater-cooled copper anodes have been widely used in experhens
work. Figure 1 shows a typical plasma jet assembly. A reactor chamber may be of my
confipuration desired to accomodate different feedinp: and quenching devices.
PLASMA tTET REACTIONS
GAS-CAS PEACTIONS TO PRODUCE A CAS AND CAS
DFCO?'POS7TION PEACTIONS TO PFODUCF A GAS
A considerable amount of research has been done by a number of people in
the area of Dlasma jet eas-pas reactions.
pases that have been investipated.
The followinp is the gas reaction producing
I
\
\
i
x
\'
x
\
I
t
\'

31 4
,Tet.
- ~2H2
CK4 (or other hvdrocarbons) Ar ,?e?
CTq + !I7--+ F3 + C=31!r2 (6) 'I
Srs + h'2+!'F3
t:P3 or CU,, crackinp rate studies (8)
In the oast several vears considerable effort has been directed to the \
investipation of the nroduction of acetylene from hvdrocarbons (reaction 1).
"he Drnduction of acrtvlene hv reactinp Fethane in the flame cf an arpon
!
olasms iet vielded an R0% conversfan to acetvlene (1).
Almost all of the methane was
converted to acetvlene and hvdrogen with little formation of soot. Pisure 7 shows
the Dower consumption VS. the feed ratio of arpon to methane. This ratio is the
most imnortant naraveter in the acetylene yield. "he minimun power consurnntion 60
Kwhr/100 CU. ft. acetvlene nroduced, corresDonded to a ratio of arpon to acetylene of
0.3. Danon and White (2) selected the manufacture of acetylene as am aoplicaticn
for plasma processinp which mipht be of interest to the petroleum industry. Pethane
was one of the pases nroposed with the use of recvcle procedure. Anderson and Case
studied the methane decomDosition reaction and compared it to available themodvnamic
data. In these exaeriments a hvdropen plasma torch was used coupled to a reaction
chamber and water ouench svstem. ?he hot hydrogen stream emitted from the Dlasma jet, \
entered the reaction chamber and mixed with a methane feed. The
after affluinp from the reaction chamher and water auench svstem.
conditions for methane produced a 30? yield of acetylene.
In a report of the National Academv of Sciences the investigation by the
Linde Company of the production of acetylene usinp a plasma jet and natural pas was
reported. This process is said to have a more efficient transfer energy to the feed
stream than does the open arc process used in Germany.
Considerable research has been done on the methane-nitrogen and methaneammonia
reactions (Tactions 7 t 3) to produce hvdroeen cyanide and as a bv product
acetvlene. Leutner 6 6)reported up to 50% conversions were obtained based on
carbon input (as methane) bv usinR either nitropen, arEon or nitroEen-argon mixtures
as the plasma eas. fipre 3 shows a schematic of the apparatus used
These experiments shoved that up to 75% of the carbon inDut as methane
WN and acetylene for reaction 3 and 90% for reaction P. Ho other hydrocarbons beside!
acetylene were found and cvanopen was present in only trace amounts.
f
Damon and Vhite (7) DroDosed the production of reducer gas by reaction 4
Using natural eas or DroDane as a hvdrocarbon source. The proDosed process=for stem+
methane refonninR would ooerate at a t emperature of 3000 to 60OOOF and Drovide a hi@
temperature reducing gas for metals and other hip.h temperature processes.
The fixation of nitropen (reaction 5) has been one of the major aDnlications/
for arc induced reactions in the
oxypen-nitro~en mixtures has beep
input has been converted to NP (6
\ !
f
q
"i
\
1
I
\
6
\

J
#
I
/
4)
I
. .
'8
r4
0
r(
P

W
v)
a
m
0
0
0
I
e
z
v,
a
?
an
a
0
I- O
W
J
J
0

317
i Cuencb svsten w.nnar;itus use8 has been full11 characteri::ec! and shown to
i be R verv COO(! Fit f ~ a r d;ifrlsi.onal -n*el.
/
DeceatlV iny~e~tj,patidns at the Pesiarch Institute of TemDle IJniversi.ty
5ave shown that hvdro sulfide ccn be bvnthesiz.e(! fron its element, hvdropen and
sulfur nowder fed in a heliur plasm ;et (12). Cnnversions as hish as 37: hased on
/ thk sulfur innut have heen o?t?ined. riuure 4 sh&is hojh the ? conversion and the . '
F/Y.whr of hyrdropcr siilf-ide fo6ed vs. Kwhr/p @f,kUlfur,iGput. As can be seen from the
fipure the raximw conversion ?, dces not necessarily; hi,ve the maxivum nroduction
eff ic ienclr.
P consic2erable nurhei of, dvntheqi.ses ha& he& .carried out usinp solid carbon
j powder or rranhite eleyent,s .as .a ,carhcn' soprc6 for reactj:ons with various materials
includj.np h-ydroren , nitrocen , .hydroben-n'itrc%n and ammonia. reactions 1 thro~wh 5
2 show the various producti.ons ,obtaine$ .hqi ihese reacti&s., Tn the case of acetylene
. ' synthesjs (reaction 2) tt1.e h~.~~&t,ylq;d obtained bv pirect svnthesis from the
elerents was 339 (11.
for recction 11 have been obtained (5). . t complete studv of the synthesis of cyanopen
\ fro* i.ts elenents was mrle tv !,eutne$-(K 6 13) .and,this-reaction Pave 15% conversion
, ' hased on the carbon innut at the ojtilrun reaction conditions.
v'
.. . .. I
r
.. 1
I+dropen cyanide yields up to 514; for reaction 3 and UD to 399
Decently Craves , Yawa and 1iiteshu.e (I4) reborted investipations wine hituninous
feE. into an aroon plasma. iet .,. .,Acqtvlene, , the princinal Drodiict , was obtained
in vields of 15 wt. ?. -his work .s'tudie< the ,effects of coal feed rate, particle
size and nlasna terperature on vieid- and nroducts' forned.
.. . . , . - . , . . - .. .
PAS-GAS ~~ACST.CIIS .TO .PAODIJC.C. A m r n
Ilvdrccarbons 4 C (2)
t'arnisch, Yemer and Cciailu: .(15) rknorted the preparation of titanjum
' nitride (reaction 1) fror titanium tetrachloride pas with either nlasna

318
FIGURE 4
I

epcrte? the noss~kil.itv of nroduct-r carton bla ck fror hvdrocarbons usinp a nlnsna
'et. !!ethane or other hvtrccarbons which would h e introduced intc the plasma flane,
uoul? be cracke2 usir.;. tvCro.ren as the ooeratinr vas an? Droducirp carbon black.
LicciC as i.re11 3s' Fzseous h-:?rncar!mns car te cse? 5s a source for carbon and the
'.'itrc khoratories (4) have exnerinente? ;iith carton Slack production frnv linuic!
h:idrocarhons.
'P"?O5 + !Z1 j Ta (8)
\!C3 t 4 :.: ' (9)
~ 1 3 + 0 Y~/CE,,+ ~
PI (10)
re0 + U7+ ?e (11)
The nroduction of metal nitrides (F! f* 17) from the elements has been
invest ipated for three elements, titanium, -arrnesiun and tunpsten (reactions 1 to 3).
The Droduction of titaniun nitride in the 1000. vield was accovplished bv usinp 200
mesh titaniur Dowder fed into a nitrooen Dlasna ;et. The titanium nitride particle
size was found to be 0.75 to 7.5 microns and the nroduct was also formed in larpe,
corpact, Folden .rellov crvstals. In a like manner tunpsten nitride was formed in
75p yield. 40% conversion to rnapnesiun nitride was obtained when nacnesium was
fed into a nitropen olasma jet.
The preDaration of'cetal carbiles has also been reDorted (12 6 17). Tipure
5 shovs the 9 conversion VS. Kwhr/a tunasten input for reaction 4. As can be seen,
the XC conversion is directlv Dronortional to the power input level. The hiRhest
conversions obtained were 43? for \,!2r and 11% for %'C. reaction 5 aSove is shown in
riplire 6 uhere .9. conversion is plotter] VS. Kwhr/v !XI3 innut. The three products of
the reaction, tunrsten (43 to Rl? cnnversicn), tunester. carhide (4 to 11% oonversion)
and ditunesten carbide (9 to 35% conversion) are fomad in a total conversion of 81
to 9UQ.. ?e naior Droduct is tiinosten which is'favored at higher Kwhr/p WOg inputs.
The reaction of tantalum and methane in a helium nlasma jet is shown in Fipure 7 where
a water-cooled cuenchinp nroba was ~ l ~ c at e d 1/3" and 5" below the plasma jet. The
effect of the ruenchinr distance is dranatic. The annunt of Ta2C formed is notapDrecfahlv
different in either case, however, the TaC vield chanped considerably bv
the n1acerrer.t of the ouenchinp device. Conversions UD to 72% have been produced
by this reaction. Tipure e shows a nlot of the tantalur-pentoxide plus nethane
reaction carried o ut in a helitic Dlasma jet. The % conversion is plotted vs. Kwhr/y
tantalun nentoxide input for two different ouenchinp nrobe distances frm the plasma
iet. Jn the case where the auencher was 1/7" 5elow the jet the Droduction of TaC

0 0 0 0'0
\ D M . S ( C ) h l l l ;
c
3
H c
L
I
\
\
t
?

1'
b
t
+

4
10
I,

323
....... -
. . . . . . . . . . . . . . . ...........
i
. .
. . .
. . . . . . . . . . . . . ................ -., ......................

324
5
Toes u? linearallv with the Kihr/p input. Where the Quenching distance was 5" a
Deak was obtained which shows that adequate quenchinrr does not take place bevond a
value of 0.4 for Kwhr/p. Note that in the second case the formation of Ta2C also
peaks and then Falls off rapidly, in contrast to the 1/2" distance. Tantalum metal
is the favored product in the 5" case.
and 18% %a for both cases.
Maximum conversions are 24% TaC, 17% Ta2C
The reduction of tantalum pentoxide with hvdroEen (12) in 7 helium plasma jet
to produce tantalum metal (reaction A) is shown in Pipure 9. P.@n the ? conversion
is plotted vs. Kwhr/p Ta2O5 input for two different quenching distances. As can be
seen the more rapid quenching (1/2". case) gives the maximum conversion (42%).
which peaks at 0.35 Kwhr/g Tap05.
In a similar manner the reduction of tunpsten trioxide was carried out in a
)
'1
\
'
helium plasma jet (17) with the quenching device 5" below the plasma jet. Conversions
as hiKh as 959, were obtained carryinp the tunesten trioxide in hydroaen into the
(
flame of the 'et. The reduction of other metal oxides has also bean experimentally
investigated 117). Ferric oxide was reduced to iron metal in a 100% conversion '\
using a helium plasma jet and carrvinp the ferric oxide in hvdroEen (reaction 11).
Titanium dioxide and zirconium dioxide reductions were also attempted hv the same
method, however, no reduction was obtained in either case. The reduction of aluminum i ,
oxide with hydropen in a helium plasma jet produced only a 2 to 5% conversion to
aluminum metal usinp several different quenching methods.
Pydrocarbons C2H2
LIOUIDIGAS FEACTIONS PRODUCING A GAS
The aroduction of ozone b means of a plasma jet was accomplished by feeding,
liquid oxyEen into a helium jet (1* v , Figure 10 shows the schematic of the apparatus
used and Fipure 11 shows the effect of liquid oxypen flow on ozone production under
constant arc conditions. The liquid oxygen acts as both a reactant and quenching mediu
Another example of this type reactant is the decomposition of hydrocarbons into
acetylene by use of a plasma jet device. Thermodynamics Corporation (19) has proved \
the feasibility of producinp acetylene from kerosene using a plasma torch. Preliminary
runs gave yields of 18% acetylene.
II II
LIOUID-GAS REACTIONS TO PRODUCE A LIQUID
\
1

1
\
i
I i
,
/
/
/
8
3
1
325
To205 +5H2 -2Ta + 5H20
50
40
30
2
0 20
W
>
z
0
V
,\" 10
0
IN HELIUM PLASMA JET
5 INCHES
0.1 0.1 0.2 0.3 0.4
KWHR. IC. To2 O5 INPUT

pl

- .-
-
- . - - . . . - .
327
n
+
c
.r
II . J
&
.r
7
.l
3
C
c
LL
z
c
-1
.-.

Under propran carried on at the.Research Institute of Temple Uni ersity for
the Clidder. Company the reactions Of terpenes in a plasma jet were studied r ?O). The
reactions shown above were investipated by use of the apparatus showed schematically
in ripure 13. A 1 1 experimental data were obtained usinr a helium plasma jet where the
terpene was added in a liquid state into the helium plasma flame. The products were
analyzed by rems of chranatipraphic absorption. The most productive synthesis was the
conversion of bpinene to mvrcene in 26% yield (reaction 1). Although this is a normal'
Dyrolsis product of A-pinene it is the first time that a complicated'molecuje has been \
croducsd. hy yeans of a nlasma iet. Other reactions included the preparation of limonene,
frorn ,>-ainene in 1% conversion a nd para-cymene
yield fmr? ;y-terninene (reactions 2 to Q).
in 2 1/29. yield and *-pinene in 2 1/2%
SUMMARY -
The cheTica1 reactions discussed above summarize the synthesises that have
been accorolished thus far using a plasma jet device and are by no means all the
synthesises studied using a plasma iet. The plasma jet has shown itself to be a useful'
tool in the area of synthesis of corpounds and has moved from the preparation of simple
materials to more canplex ones recently, With the ever increasing number of investiga- \
tions heing carried out by private industry, there is no doubt that the plasna let
will becone a cmnercial chemical nmcess device. Its potential has been just touched
and with each new use a whole field of investipation is onencd. \.
,

'
POWER
PLASMA GAS
l e , '
I ! I
I
329
- PLASMA JET GENERATOR
j-~ &TERPENE IN
! !
I 1 /- OUENCHING CHAMBER
/ I
.
4
i
1
I
'
:.
-
:
j
:
1- H20 IN
I '
I
I r- ' PLASTIC
I
I
I
1 BAG
+LIQUID
COLLECTOR
I

330
PPTRFKCCS
1. Leutner, 1'. l!. & Stokes, C. c. , Ind. Pnp. Chem. 2 341 (1961)
2.
3.
4.
Damon, F.A. F, thite, D.E., "roposec! "lasma Chemical Processes of Interest to the
Petroleurr Industrv, Fenort KO. PLR-115, Plasmadyne Corp. , Santa Pna, Calif.
(Feb. 8, 1962)
I \
Anderson, .l.E. E Case, L.K., I & F.C. Procoss Desipn and Develop. i(1962)
Development and Possible Applications of Plasma and Related High Temperature
. Ceneratinp: Uevices, Deport !JPP-l67-!! , Pivision of Jhpi.neerinp and Tndustrial \
Research , Hational Pcadeav of
(PUP. 30, 1960)
5. I,eutner, F.W.
6.
I.
R.
9.
Sciences , Fational Research Council , l?'ashineton. D.C.
I .E F.P., Process Design and Develop. 2 315 (1963)
Crosse , /I..V., Leutner, H.W. &. Stokes, C.S. , Plasma iTet Chemistry, First Pnnual Feport
The Research Institute of Tenple University,. Office of :!aval Reseerch Contract
K!':!?-3085(02), (Dec. 31, 1961)
DaRon, R.A. & G!hite, D.F., Typical InorFanic and Inorganic Miph Temperature plasma '\
,?et Reactions, Peport ?!o, PLR-119, Plasmadvne Corp. , Santa Ana, Calif. (Sept. 20, 196:
Stokes, C.S. F, Knipe, W.H., Ind. rng. Chem. 52, 287 (1960)
1
Office of Naval Research
Grosse, P.V., Stokes, C .S., Cahill, J.A. & Correa, ,T..T., Plasma Jet Chemistry,
Final Feport, The Sesearch Institute of Temple Universitv,
Contract NONR-3085(02) ,(dune 30, 1963)
10. Bronfin, D.?. & Hazlett, P.H., I & C.C., Fundamentals 2 472 (1966)
\1
. 11. Freeman, H.P. & Skrivan, ,?.I". , A .1 .Ch.E. Journal 5 450 (1962) c
12. Stokes, C.S. & Cahill, ,T.A., Plasma Jet Chemistry, Final Report, The Research
Institute of Temple University, A i r Form Office of Scientific Research Grant
775-65, (December, 1965)
13. Leutner, R.W., Ind. Ene. Chem. Process Design &-Develop. 1166 (1962)
I&. Graves, R.D. , Kawa, W. & Hiteshue, R.W. , I t E.C. Process Desi,qn E, Develop.
15.
16.
17.
18.
19.
\
\
\
\
\I
'.
f
s'
59 (1 6
Parmisch, F., Kevmer, C. & Schallus, E., Anaew, Chem. Intemaut.' Edit. 2 238 (1963) \
Carbon-Elack Formation Usinp the Plasma Flame in Yicro-Chemical Reactions, Plasma-Fax,
Bulletin PF-3, Thermal Dynamics Corp. Lebanon, K.H. (Oct. 1960)
k
Stokes, C.S., Cahill, J.A., Correa, J.J. & Grosse, P.V., Dlasma Jet Chemistry, Final
Report, ?he Research Institute of Temple University, Air Force Office of Scientific
Research, Grant 62-196 (Dec. 1964)
Stokes, C.S. & Strenp, T,.A., I & L.C. Product Research & Develop. 5 36 (1965)
Plasma Flame Used in Formation of Acetylene from Kerosene, Plasma-Fax Bulletin
Pr-2, Thermal Dynanic CorD., Lebanon, N.H. (kt. 1960) c
50. Stokes, C.S. & Correa, ,?.J., Terpene Reaction in a Plasma Jet. Final Report. The
Pesearch Institute of Temple Univ. The clidden Company (spnsor)(Dec, 19610 \
-
?

331
f'ENI?AL PF'FCLCCFC
Dernis, P.R., qmith, r.D., Cates, n.1'. F. Sond, T.O., Dlasna tTet Technolow
'.ASP CD-5033, "ct. 1965
Stokes, C.C., Chenical Peactinns with the plasma Jet, Chev. Cnp. PDril 17, 1965,
D. 191
tlulburt, I..)!. & Freeman, K.O., Chepjcal "eactions in the Plasma tTet, Trans. V.Y.
Pcad. of Sciences E 770 (1963)
5taff Feature, Plasma - Fourth State of Vatter, Ind. I'np.. Chem. 55 16 (1963')
$lohn, R.R. t nade, W.L. rlecent Advances in Flectric Arc Dlasna Generator Technology
APS .Taurnal
Creenlee, R.L., '"he Dlasma ,'et in Chemical Processinp. Clyde hrilliams & Co.,
Columbus, Chi0 ("arch 1, 19631, Creenlee, R.E. & Fickley, W.H., Potential
Pcplications for plasma %'et Drocessinp in the petrochemical Industry, Clyde
Villiams & CO., Coiumbus, Ohio (April 30, 1963)
Peed, 7 .Pa , Dlasma "orches , International ccience and Technolop, June, 1962, p. 42 '
Parynawski, C.W., Phillips, V.C., Phillips, tT.R. F, I'iester, N.K., Thermodynamic of
Selected Chenical Systems Potentially Applicable to Plasma .let Synthesis, I & E.C.
Fundmentals 157 (1967)
I

INTRODUCI'ION
332
PRODUCl'ION OF BYDROGEN CYANIDE FROM METHANE I N A NITROGEN PLASMA JET
I. Reactive Species Titration; -her Q,uantitative Studies
Hark P. Freeman
Central Research Division, American Cyanad Co., Stamford, Cmnecticut
The reaction of nthane vith a nitrogen plasma to make hydrogen cyanide and acet-
ylene is of considerable interest. Conversions are high enough t o be of commercial
interest on the one hand,(l,2) vhile the formation of H a in particular proceeds in
(1)
(2)
E. M. Hulburt and M. P. Freeman, Trans. N.Y. Acad. Sci., 2, No. 25, 770 (1963).
E. W. Leutner, Ind. Ehg. Chem. Process Design Develop., 2: 315 (1963).
such an interesting and reproducible way that clarification of the details of the
reaction should considerably advance the use of the plesma jet in 8ynthOtiC chew
istry, and further might be expected to contribute rlgniflcantly to baaic chendcal
knowledge.
The forrmlly identical synthesis of BCB from "active Ditrogon" aad methane has
been extensively studied(~5) for 'mre than a half century. %at the mysteam are
(3) K. R. Jenninge and J. Y. Llnnett, Quart. Rev. (London) l2, 116 (1958).
(4) G. G. Manella, Chem. Rev. 63, 1 (1963).
(5) 1. E. V. Ewms, C. R. Freeman, and C. A. Winkler, Can. J. Chem. 2, 1271
(1956).
different is apparent, for the high-voltage discharge Is Z high excitation &rice,
whereas the plasma jet is thought to be nearly in local therml equillbrlum .nd
hence a l w excitation device (spectroscopically speaking interredlate betveen 'arc
and spark.(6)) Furthermore, the plasaa jet experiments an perforad at one-half at-
(6) F. A. Kovoler and Yu. K. Kvaratskheli, Opt. Spectr., (IkSR) (Eugllgh Trml.),
- io, 200 (1961).
I

i
J
333
; mosphere (as opposed to about 1 torr) and at an average temperature tventy times
as high on the absolute scale as room temperature, vhere the bulk of active nitrogen
experiments have been performed. Finally, in the plasma jet the carbonaceous
species reacting is evidentially not methane. That is, the products other
than HCN are acetylene and higher acetylenes vith various degrees of saturation.
These are the same products that would form if the jet vere, say, argon. It has
b been shown elsevhere(7) that the precursors for these products form rapidly compared
I
/ (7) M. P. Freeman and J. F. Skrivau, A.1.Ch.E. (Am. Inst. Chem. Engrs.) J., g,
450 (1962).
I
>
,' to the time for mixing of the methane vith the jet.
I
,'
There is no question that the plasma Jet provides a difficult environment in which
to do "good vork" in the usual sense. The enonnous temperature gradients and consequent
inhomogeneities are generally thought of as being a sort of physical chemical
bar sinister. On the other hand, perhaps because of their extreme magnitude, the
i effects due to the temperature gradients are found to be sufficiently reproducible
to allow systematic investigation of the jet as a whole, vhich in its hotter parts re-
' presents in a steady state flow situation a chemically unique environment found only
/ in high intensity arc devices. As ccsnpensation for tackling this difficult environ-
ment, the investigator need not vork with trace quantities, but instead vith partial
pressures and relative conversions at least tvo orders of magnitude higher than those
I encountered in active nitrogen research.
) The vork reported here represents a systematic continuation of a study performed
some t ime ago.(8) The procedure followed then 8s nov is to add methane through an an-
I
* (8) M. P. Freeman, "The Nature and Quantitative Determination of the Reactive
$'
5 /
'
Species In A Nitrogen Plasma Jet," presented at the 147th Natl. Meeting, Amer.
Chem. SOC., April 5-10, 1964, Philadelphia, Penn.
4 calibrated flov rate and a measured heat flow). It has been shovn that under these
\ condition xing is very rapid, as is the drop in temperature deflncd from average
nular slot to a confined nitrogen jet of precisely deflned average enthalpy (i.e., a
enthalpy. 17Y Aiter about one millisecond the resulting flow of high temperature species
t/ is further chilled by the entrainment of cold product gas (the fastest of several
\ quench methods investigated on the basis of its efficacy in quenching the ammonia
decomposition reaction)(T) and the flov of HCN in the product gas is chemically de-
f termined. Except as noted, the data are taken at 350 2 20 torr chamber pressure as
1 this pressure is found to reduce the formation of solid product to an insigniflcant
) level. As the methane is added at various flw rates the corresponding rate of production
of HCN is noted.
{
'
Just 88 in e.g., Winkler's work vith active nitrogen, a
plateau is observed vhich seem to indicate that some active species is indeed being
titrated (Figure 1). Of utmost interest in'the older study vas the provocatively
simple dependence of the plateau level on jet power level (Figure 2). Because this
may in turn be shown to correlate with the rate of production of ions in the arc
!

335
I I I I I I I I 1
1
0.12 - -
- -
y 0.08 -
1.
Y
.t
- -
0 0
0
-O- -
I I I I I I
0 0.2 0.4
%CH4/G
0.6 0.8 1 .O
?

; it raised the question, still &?zesolved, as to whether the ions might
’ not in some way be responsible for the chemistry. As a direct consequence there is
a long-term quantitative spectroscopic study of plasma jets(9) currently in progress
/(9) M. P. Freeman, “A Quantitative Examination of the LTE Condition in the Ef-
1 fluent of an Atmospheric Pressure Argon Plasma Jet,“ CE-JILA-ONR Symposium
on the “Interdisciplinary Aspects of Radiative Energy Transfer,”. Philadelphia,
Penn., Feb. 24-26, 1966. In Press.
/
/
J’
that is expected to ultimately yield information on the nature and quantity of ions,
atoms and high temperature molecular species flowing in jets of common plasma ma-
terials.
Although thermodynamic calculations had previously been performed for the nitro-
gen-carbon-hydrogen system(l0) they were not in a form easy to compare with plasma
/
‘ (10) C. W. Marynmski, R. C. Phillips, J. C. Phillips, and N. K. Hiester, Ind.
Eng. Chem. Fundamentals, 1, 52 (1962).
/
1 jet results obtained under normal operating constraints. The calculations were there-
; ,fore reproduced(11) with the pertinent parameters varied to conform to the exigencies of
;I
i
1 (11) B. R. Bronfin, V. N. DiStefano, M. P: Freeman, and R. N. Hazlett, “Thermo-
\
1
chemical Equilibrium in the Carbon-Hydrogen-Nitrogen System at Very High Tem-
peratures,” Presented at the 15th CIC Chem. Engr. Conf., Q,uebec City, Quebec,
Oct . 25-27, 1965.
7
j
,/ plasma jet operation. Figures 3 and 4 show the results obtained when solid carbon
is suppressed in the calculations to be consistent with its absence in the observed
I/ products. An important fact immediately emerges. Neither at any experimentally
/ attained average enthalpy, nor at any experimentally used ratio of methane to nitrogen,
has the thermodynamically expected yield of HCN for a well-mixed system been
(Note that the apparent fall-off in calculated yield at high power levels
/ ~ ~ ~ % methane w flow rates is due to the competitive formation of cyano, CN, which
, may be presumed to be an HCN precursor.)
7 The relationship between the observed and calculated equilibrium yield for the
CHb-N2 System is strikingly similar to that recently reported for the F-C-N system by
1 aronfin(l2) who thereupon advanced the plausible hypothesis that in either case the
I
r‘
(12) B. R. Bronfin and R. N. H,azlett, Ind. Eng. Chem. F’undamentals, 5, 472 (1966).
#
i
observed yield is in some way a consequence of equilibrium considerations. Now,
from earlier work on the acetylene system(7) it would seem that the composition and

3.0
2.5
2.0
H2'N2
15
3. -
I .o
Q5
'0 02 0.4 0.6 OB u) L2 1.4 16 W u)
Calculated yieldr of HCH, .c&yl.no urd 4drg.n rrlatim to
initial nitrogen renilting f'ra tho indicatd nitraeon-uthw rirhP0r.
The methane ir initially aold md the nltr0e.n initi&lly .t tho indicated
onWpy. Solid cubon hu boat mgpro8r.d u a rp.cior in thlB adslatian.
'\
'\
'\
i
?,
r
<
t
r
,
i

p .I
V
I
.=
-
- HCN/C(CALC.
WITH C (SI )
337
100
ENTRANT HEAT FLOW KCALIMOLE Nz
4. Old HCN plateau yield data plotted y~ initial nitrogen enthalpy.
Also shown are the appropriate theoreticilly calculated curves for HClf
(W aquillbrlum), HCH (cubon suppressed) and A at-. The dMhed
line indicates where the stmewhat more scattered experl,mentrl data
would have fallen if plotted y~ 2" reactor cL enthalpy.
HEAD
INTERMEDIATE
SECTION

8
enthalpy dependence are not what one would%xpect for mixing and subsequent freezing
of a reaction in a confining tube and further for the shorter reactors the enthalpy
at the exit is still very high. We shall therefore disregard this possibility here.
(Note however, that Bronfln is currently testing this nodel by means of computer
(13) B. R. Bronfln, personal cammunication.
I
i
$
I \
simulation.) The expected way in which equilibrium could control the reaction is for
the product distribution to be a function of the enthalpy and canposition profiles
at the of the reactor where there is an onset of rapid quenching. The calcula- \
tions indicate that composition is only of secondary importance in the "plateau"
region so that presumably the enthalpy proflle would be controlling. This differs
\
from true titration in the important respect that in titration one in effect irre- (
versably consumes a certain potential of the nitrogen Jet to form HCH that is clearly
a function only of the temperature and/or composition and velocity profile in the
\
nitrogen at the point of mixing but before mixing occurs. If we a~sumc the velocity \
profile of a plasma jet is uniquely determined by the enthalpy proflle,(g) then the
reacting potential as a consequence would in turn be uniquely related to the heat
flow in the gas.
1
Despite this difference, there is no way to distinguish unambiguously between these
two possibilities using but one reactor. This is because fractional heat loss from a
plasma in a particular reactor type has been shown to be primarily a function of
reactor length and arc unit design,(14) so that the ratio of exit heat flow to inlet heat \
(14) J. P. Skrivan and W. VonJaskowaky, Ind. Eng. Chem. Process Design Develop. 4,
371 (1965).
flow is nearly constant. The primary objective of the work reported here vu there-
fore to carefully distinguish between these possibilities by performing identical
titration experiments in tvo or =re reactore that differ signiflcantly in length in
order to determine unambiguously whether inlet or exit heat flaw controls the
reaction.
It had been inferred fran the earlier work by rather incomplete evidence that smhow
the capacity to make HCN is primarily dependent on arc conditions no matter hav
far removed the injection point is fran the arc unit itself. A further objective of
this experiment wat# therefore to systematically check this tentative conclusion by
careful control and variation of the injection point on a losg nactor. Ran the
standpoint of the titration hypothesis, if the arc unit conditions are indeed controlling,
then a very long-lived reactive species is Implied; such a species is hardly to
be expected under these experimental conditions.
Finally, at the same time the older vork VY being done, Leutner(l5) using a very
short tubular reactor of othervise the sam design found he vu able to work at
(15) E. H. kutner, Ind. Eng. men. Procesm Design Develop., 2, 3 5 (1963).
,

1
1
339
atmospheric ressure and achieve a nitrogen fixation of 12.5% (erroneously reported
as 21.9WP in a nitrogen (62%)-argon jet (erroneously reported as pure nitrogen)
(16) C. S. Stokes, personal co&unication.
.J with a total flow of 5.0 l(STP)/min. and a power flowing in the gas of 11.5 kw X
55%(16) = 6.32 kw. It was deemed desireable to reproduce his reactor as nearly as possible
to attempt to see if the two sets of results were consistent. Such compari-
'
son will be made in Part I1 of this paper, where the effect of argon dilution
of the plasma will be discussed in some detail. The contribution such a reactor
makes to the present work is of course to extend the range of reactor lengths studied.
J
EXPERIMENTAL
1
/ Apparatus
i
(
1
3
Plasma-jet reactors consist of three parts, head or arc unit, intermediate section,
and quenching section. The plasma-jet heat unit used for this study is a Thermal Dynamics
640 Plasma-jet with "turbulent nitrogen" electrodes, powered by two 12 kw welding
power supplies, open circuit voltage 160 volts, connected in parallel but with
opposite phase rotations on their 3 $ input so as to minimize line frequency ripple in
the output. The intermediate sections (Figure 5) are made of copper and are fully
water-cooled, as is the head. The three intermediate sections are themselves modular.
Of length 2". 2", and b", they are mutually compatible and can be joined in any order
to make a reactor of length 2, 4, 6, or 8 inches with a feed port at any multiple of
2 inches. Note that the feed rings are not exactly reproducible in that there are
residual gaps left when the surrounding gaskets are tightly compressed by the joining
threaded parts. The "Leutner reactor" standard Thermal Dynamics spray nozzle with the
solids injection port, which is about l/h" from the nozzle exit, opened up to a 180'
slot.
The spray nozzle and the turbulent nitrogen electrode used with the intermediate
reactors has a 7/32" diameter as do the intermediate reactors. Insofar as the various
reactor configurations vary only in length and feed point, it w ill suffice to distin-
guish between them with a bracket specifying first the distance from the point of heat
balance to the point of methane feed, and second the distance from the point of heat
balance to the exit of the reactor. Thus the 8" reactor fed at the 2" point would be
designated (2"; B"), vhile the Leutner reactor is (-1/4"; 0").
The quenching section where the hot stream of plasma and reaction products are quenched
by entrainment of cold product gas is simply a stainless steel pot 11 inches long and
11 inches in diameter sparsely wound with soldered-on copper tubing. All parts subject
to heat damage are vell-cooled, but between the windings the pot may get hot enough to
cause flesh bums. At the outlet of the quenching section is six feet of 1 inch thick
rubberized acid hose. This in turn is fastened to the bottom of a vertical mixing section
consisting of a three foot long 2" diameter pipe loosely packed with glass wool.
The carbon dioxide is mixed with the product stream at the inlet to the mixing section.
The top of the mixing section is connected to a high capacity steam vacuum jet with an
automatic control valve for maintaining desired pressures.
A Toeppler pump is arranged to withdraw 522 ml of gas from the top of the mixing section
at room temperature and at the reactor pressure. This aliquot is then collected
for analysis in a suitable gas collection system.

340 .
Gas flaws except for methane are metered by orifice gages calibrated by water dis-
1
placement to within 1% for CQp and N2, respectively. Methane flow, much less critical,
is determined by a rotameter calibrated by calculation. All cooling water flows are
i
determined by experimentally calibrated rotameters. Cooling water temperature rise
is determined by suitably graduated, interconsistent mercury thermometers.
Procedure
Heat flaw in the nitrogen plasma at the point of methane introduction is determined
by subtracting from the voltage-current product in the arc the heat ldst to all cooling
water supplies up to that point. For the most part, just the heat fldng at the
exit of the head is required. For data taken at a particular heat flow an attempt is
made to keep the heat flaw constant. In this endeavor the relatively great intrinsic
stability of plasma jets made by this manufacturer help, but especially in runs lasting
for several hours it is necessary to continually introduce small corrections. TNs
control is greatly facilitated thmugh use of an analog camputer that continuourly
monitors voltage, current, and cooling water temperature rise and either directly controis
the rectifiers or displayo the net heat flow in killowatts continuously on a recorder
chart so that manual corrections mag be introdvced as needed. Heat levels
given are generally correct to within 25%.
Except where noted, the pressure in the quenching chamber is kept at 350 2 20 torr.
. The actual pressure of each sample is known to fl torr but that is not a significant
datum in the analysis and is used only as a consistency check. Quench section pressure
measures intermediate section pressures fairly vell, but probably not the arc
pressures because of the pressure drop through the front orifice of the plasma jet.
These arc units are not instrumented to measure the pressure inside the head. J
Whenever methane flow rate, paver level and/or pressure conditions are changed, the
system is operated for eight minutes before taking a sample. This is found to be
sufficient time to establish a constant composition.
The collected gas aliquot is slowly bubbled through 200 ml of ice cold-caustic con- ,
The half liter space over the caustic is initially
1
taining 12.5 millimoles of base.
evacuated so that the entire sample, together vith the air used to flvsh the lines,
might be collected in the caustic and the space over it. This is followed by one minute
of Vigorous shaking. This procedure has been found satisfactory for the quantitative
recovery of Cog and HCR. Total acid insthe gam is then determinod by back-titration
with 0.500 HC1 until all the carbonate haa been converted to bicarbonate (pH = 8.3). 1
Ammoniacal M is then added aa an indicator and cyanide determined by precipitametflc ,
titration with 0.0100 J silver ion. This permits the initial ratio of ECIi to CO2 to
be determined. Because the absolute flaw rate of Co2 is known, the absolute production
rate of HCN follavs directly. Hote that for convenience in preeentation the flow rate \
of HCR is nlwws presented as some fraction of 0.383 l(STP)/sec.(0.0171 gram moles
sec-1) so that it mey conveniently be compared to the most often used flw rate of nitrd

RESULTS 341
Composition Dependence
Figure 6 shows typical "titration" curves for the different situations of interest.
In every case a plot of HCN produced vs. methane added (both normalized to the "standard
flow rate" of 0.0171 l(STP)/sec) divides cleanly into two renimes separated by a
break to which we refer as the "equivalence point." To the left of the equivalence
1 point, the yield is simply dependent on methane feed rate but not on the power level.
1 Lines Of slope 1/3 and 2/3 are included on the graph to facilitate intercomparison in
'
)
\
this region.
Although the data are too scattered to draw firm conclusions, it is clear
that the break quite generally occurs along the line correspondine to a slope of 113.
In this region it is as though the nitrogen were somehow present in excess. To the
right Of the break the HCN throughput is seen to depend only veakly on methane but, as
is demonstrated below, is a strong and simple function of heat flow in the nitrogen
before mixing. Excluding for the moment the (-1/4"; 0") data (Figure 6-d), the methane
dependence on the right shows a real linear increase with methane flow rate in
every case, but this slope is not found to be particularly reproducible since it is apt
to change slightly when the apparatus is demounted and reassembled. Hence the scale
has been chosen to emphasize the plateau-like character of the curve. Note that the
data excepted may be explained by the fact that in this case the exit rather than the
lentrant heat flow is regulated. Because of the small heat loss in this short section
only a second order effect of this magnitude would be expected.
Power Level Dependence
To take account of the residual slope of the "titration" curves, intercomparison of
power-level-dependence studies is done at the same methane flow rate corresponding to
one half the "standard" flow whether it is an interpolated point taken from a full
titration curve as shown in Figure 6 (filled points) or an isolated measurement (open
points). The best line through the oid data (Figure 2) appears in Figures 7 through 9
as a reference to aid in intercomparison.
Note that Figure 6-a should be exactly con-
sistent with the older data of Figure 1 while the data of Figure 7-b should exactly
reproduce Figure 2. The extent that they fail to do this is a fair measure of the
nonreproducibility of this experiment when performed in different laboratories, with
different equipment, vith the chemical analyses performed by different persons.
In Figure 7-b data from two different 2" reactors are intermingled as shown. This
is to be compared with the (0"; 8") data of Fieure 7-a. It is quite clear that within
the experimental scatter it would indeed be difficult to improve the agreement.
Now the heat leaving the 8" reactor for a given input is only about one half that
leaving a 2" reactor for the same input, so that there can be no question but that the
HCN production depends only on the heat flow at the point of mixing (or at the exit Of
the head) and not at all on that at the exit of the reactor.
Mixing Point Dependence
Figure 8 demonstrates clearly that the ability of the nitrogen to make HCN does not
persist d m the tube at its high initial level, but rather decays as the heat flowing
in the gas decays. Shown are a particular set of (6"; 8") data plotted on the left
- VB heat flow at the exit of the head, and on the right as a function of the heat flow-
ing at the point of mixing, the 6" point. Hence that part of the earlier work indicating
the existence of a "long-lived" reactive species is wrong.
Leutner Reactor
Data for the simulated Leutner reactor are shown in Fi&ure 9. Despite the fact that
these data were taken at atmospheric pressure it is clear that the results are completely
consistent with those of the other reactors. This atzrees with an observation
made in the earlier work that there is but slight pressuke dependence for this reaction.
.._ . _,.-. ..
., .... ..~. - ...
'

*
L 1 1 1 1 I I 1 1 I I I
U
-
0.10 1 / 6099 kw. (0";8") -
0.Q) - I /
0.09
0.m
0.07
u
6 0.06
X
.e - 0.05
9
2 0.04
p 0.03
0.02
O.O1 0 L
342
-
QW
1 1 1 1 1
.I
-
METHANE FLOW (hCHJC)
- o '0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 a9 1.0
' ATMOSPHERIC
l/l I I I I 1 1
4770k~(-1/4";0")
PRESSURE 1
- rn A .
-
- " -
1 1 1 1 1 ' 1 1
6. Typical titratiaa curre. for the cue notad. nomd.i;ml productim
rate or HCR V. flow rat. or m e . DMII.~ liar. or ~iape i/3 .nd 2/3
ue inciw~~-?or referonce.
1
0 I 2 3 4 5 6 7 8 9 IO
HEAT FLOll (kw)
-
Y
0.09 I
0.08
0.07
3 0.06
I
'5
0.05
2
0.04
5) 0.03
0.02
0.01 1
-
!
I
~

0.09
0.08
0.07
9 0.06
2
U
-
.$ 0.05
0
0.04
>
2
U
I 0.03
0.02
0.01
0
0.13
0.12
0.1 I
- 0.10
W
\
z 0.09
U
0.08
v
2 0.07
ul
;
0.06
'5 0.05
I
0.04
. 0.03
-
-
-
-
-
-
343
1 1 1 1 1 1 1 1 1 1
( - I /4 'I ; 0 'I)
0.09
0.08 -
0.07 -
0.06 -
0.05 -
1 I
(b)
I I 1 I
-
0.04 - -
0.03 - -
0.02 - -
0.01 -
-
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6
ENTRANT HEAT FLOW (kw) , FEED POINT HEAT FLOW (kw)
8. Aorrallzed production rate of HCN fo- the eight inch reactor fed
methane at the sFx inch point, (6"; 8") data, v8 heat nm at (a)
the arit of the hend end (b) at the 6" feed poGt. Fill.& points
were intarpolated trca full titration curves. The "nomd" line of
Figure 2 is included in either cam for reference.
n
i
:::: 1 , ', I , I ,' , I , ,
0
0 I 2 3 4 5 6 7 8 9 1011
HEAT FLOW (kw)
9. Romsllxed production rate of HClO in the Lautner reactor vs heat
flow at the of the reactor. The "noxmd" curve of rigur2 is
included for reference.

344
Because of the use of diluent argon in his vork, comparison with Leutner's results must
await Part I1 of this paper which will explicitly treat this complication. OS significance
here is the fact that reactors from 1/4" to 8" in length give the same result,
dependent only on entrant heat flow. (Actually for the 1/4" reactor the exit heat flow
is measured, but presumably in this case there is a nearly negligible difference.)
Product Distribution
Plateau Region - Mass spectrometer checks made on product formed i
the "plateau"
region show 12, H2, HCN, C2H2 and CHI, together with some small quantities 4 of higher a
\\
acetylenes to be the only species present to any appreciable extent. The R2 is present
of course as a solvent while the H2 is simply the balance of the hydrogen. The HCN
is apparently fo2med by some sort of irreversible process to be discuased further belov
while the relative amounts of methane and acetylene seem to conform to considerations
investigated previously(7) for the cracking of methane in an argon jet.
\
\
\
1
Initial Region - To the left of the break region where the potential of the nitrogen
Jet to react is in excess, one might expect all of the methane to be converted to
HCN, i .e. , an initial straight line of unit slope, but -such is not the caae. Earlier \
work seemed to favor an initial slope of 1/3 for all titration curves. This vould \.
imply that one mole of acetylene is formed for each mole of HCN produced. At the tima,
however, admittedly crude mass spectrometer checks shoved no more than two-thirds to
three-quarters mole of acetylene to each mole of HCN. The work reported here indicates
an initial slope corresponding to one-quarter to one half mole of acetylene for each
mole of HCH. At the breakpoint, however, equimolar quantities of Ha and acetylene
would still seem to be the rule. The analytical md sampling apparatuses were neither '
designed for accuracy nor high precision in this low yield region and it is possible
that the differences in the low HCN yield region might have reflected some small change
in analytical procedure. It seema more probable, though, #at this region is indeed -
not reproducible and that product distribution here reflect8 some intangible of the
process such as "mixing efficiency", etc. (Note that the reactors are constructed 80
that the widths of the slots through vhich the methane ?love art not precisely reproducible.
)
DISCUSSIOB
The Titration Currc
It is clear that the reactor length md hence the heat flw at the onset of sudden
quenching is irrelevant; the heat Zlw at the mixing point evidentially gov~ln~ the
extent of reaction. It is -her seen that the ability of the nitrogen jet to react
vith the methane decreases as the nitrogen flow6 dawn the reactor in such a vay that
its potential to react with methane, at least with this reactor geometry, is a function
only of heat flow in the nitrogen before mixing.
Initid Region - Roln a different perspective, with no methane ?laving in a long
tube the reactive potential of the nitrogen jet is seen to persist to 8- appreciable
extent almost indefinitely, but &cays an the heat flow decays, pres-4 throw
heat conduction processes. As the methane flow is introduced, 8- of this "potentid
to read" (PR for brevity) is now rued up by the methane, vhile the balance decqys by
the heat conduction process. This gives rise to the "initial regIo2l.l'
Equlrslence Point - As the =thane flow is increased, more of the PR of the nitro-
gen jet is wed by the =thane uatilthe equivrlence point is reached. At this point
nane of the PR sunires the mixing of the Jet with the methane md the break in the
cUP7C occurs.
~
Plateau Region - Because mixing is not itmtantaneoum at the point of mixing, sane
of the PR is still used up by heat conduction processes at the equivalence point. k
the =thme flw is further Increased, the equivalent amount rixes closer and claret
I
(
[
\
~

II
\ to the Slot So that less of the residual d 2s lost to heat conduction. This latter
process probably gives rise to the observed residual slope in the "plateau reuion" of
\ the titration CUES.
/
'
5
I
I
/
The Potehtial To React
It remains to propose an explanation for the "PR." Although something is being
titrated, it is by no means clear just what it is. Nor, as of the present time, has
anyone experimentally characterized a nitrogen Jet sufficiently well to clearly distinguish
between likely alternatives. Nonetheless, it is instructi& to examine some
of these possibilities for their heuristic value.
Heat Balance - From Figure 2 a fairly constant heat requirement of 1280 kcalfmole
may be obtained. This corresponds very nicely (and. probably fortuitously) to the
endothermic heat of the most probable reaction at 6000°(17):
(17) "Janaf Thermochemical Tables .I1 Dow Chemical Company, Midland, Michigan, Dee-
ember 31, 1960.
b
I
, or that of the equally probable reaction at 4500':
f
1
AH = 1281 kcal
AH = 1225 kcal
I From this point of view it would appear that the heat of the jet is in some way being
> titrated by the methane.
Active Species - One might well ask however, if it is Just the heat being titrated
why docs the reaction apparently stop when the core of the Jet is still 4000-6000'K?
And why is the amount of cyano (CN) formed so sharply limited and constrained so far
below the equilibrium value? Note that below 4000' strong exothermic reactions occur
that delay further cooling, so that fast quenching cannot be the answer.
The most
ready answer to these questions is that the HCN reaction is far too slow to equilibrate
in Jet residence times. Of the manifold complex of forward reaction paths lead-
'ing to equilibrium, only a few of them will be fast enough to produce HCN in the time
available. But these fast reaction paths might well involve nitrogenous species which
at thermal equilibrium corresponding to the average enthalpy of the jet, would flow in
trace amounts but which, aa a consequence of the hot core, are present in the plasma
jet at many orders of magnitude higher throughout. Thus we are led naturally to consider
this as the titration of some sort of especially reactive species in the jet.
Consider for example a Jet producing 0.00171 moles sec-1 of HCN. From Figure 2, we
1 I can see that this requires about 9 kw. OS heat flowing in the gas. If we assume there
f ia a small core to the jet at 12,000' (enthalpy at one-half atmosphere = 500 kcal/
L mole),(l8) then the heat flaw can be accolmted for by assuming the hot core occupies
i
1 -
4
t
I (18) F. Martinek, "Thermodynamic and Transport Properties of Gases, Liquid and Solids"
I'
J
j
3
J
I
McCraw-Hill Book Co., Inc., New York, N.Y., 1959, p. 130.

about half the dlamcter of the jet (one fourth 345 the area) and that the nitrogen flowing /
outside this core carries negligible enthalp . In this calculation is the unproven
assumption carried over from the argon jet(9 J that the mass flux of nitrogen is constant (
over the entire cross-section. Under the assumed core conditions the Set is 100%
disssociated and the atoms 17% ianized. Tfie fluxes are then:
atom flow: 1/4 X 2 X (1 - 0.17) X 0.0171 = 0.0071 moles sec-1
ion flov: 1/b X 2 X 0.17 X 0.0171 = 0.00145 moles sec-1
I
The ion flw is thus seen to be in excellent agresment vith the 0.00171 moles sec-1
production rate of HCN. To some extent this may be due to a fortunate choice of hot
core temperature but is nevertheless a promising possibility.
On the other hand there seems to be no good way to account for the reaction on the
basis of nitrogen atoms, in this case present in large excess, suggesting that the
nitrogen atoms are somehow deactivated before the carbonaceous species enters the hot
core. Insofar M the hydrogen from the dissociation of methane must completely flll
the reactor in a very short vhile, it seems probable that tha nitrogen at- u e
deactivated fram this initial inmion, and that the carbonaceous species reacWwith
the nitrogen ions (vhich might well be molecular ions by this time) or aome other
species sometime later.
CONCLUSION
By systematic variation of 'reactor geometry it has been conclusively demonstrated
that methane added through a peripheral slot to a nitrogen jet titrates, apparently
in the true meaning of the word, some potential of the nitrogen Jet to react with the
thermal decomposition products of methane to form HCN. It is further demonstrated
that the potential to react is simply related to heat flow even far down the reactor
and is therefore probably the consequence of some steady-state temperature and/or
composition proflle in a flw with local thermal equilibrium.
For their heuristic value, tvo superficially different explanations are proposed to
explain the "potential to react." On the one hand, the experimental endotherm of the
reaction at the equivalence point is ehom to be quite consistent vith the heat flov-
ing in the hot core of the jet, for a jet model consistent with vhat ve light expect.
On the other hand, for the same heat flov and jet model, the yield is shown to be con-
sistent vfth the flw rate of e.g., ions at the point of mixing and it may equilly
well be postulated that the ions or some other identifiable species are in fact an
active ingredient being titrated.
It is clearly pointless to attempt to conjecture further on t he mechanism invol-d
in the titration vithout more direct evidence on the species and relocity profllee in
the jet. A spectroscopic program currently undenqy in there laboratories dl1 pre-
sumably help to flll this gap.
Acknowledgement
Usnp people have contributed in nome vay to the success of this work, but in pu-
ticular the author wishes to express his appreciation to Mr. Charles Mentzer
(Chemical Engineering Department, Princeton University) who helped perform some of the
experiments , and to Professor Hugh Hulburt (Chemical Engineering Department, Iorthvesfern '
UnivCmity) and Dr. B~rrg R. Bronfln (United Nrcraf't Laboratory) for their invaluable
di8cunsions and continuing Interest.
1
I
I

i
4
2
HYDROCARBON-NITROGEN REACTIONS IN A TRERMAL INDUCTION PLASMA
Bary R. Bronfin
United Aircraft Research Laboratories
I East Hartford, ConneLticht 06108
I
4, Gas temperatures anove l5,OOO K have been observed in plasma generateddby
radio-frequency induction coupling'. This high thermal erlergy regirre becon?es of
iqcerest to the chemist for the investigation of highly endothermic reactions.
I?. particular, this paper will report reactions between methane and nitrogen fed
to a thermal induction plasma ar,d subsequently quenched to yield hydrogen cyanide,
acetylene and hydrogen.
/
PREVIOUS STUDIES OF THE E-C-N SYSTEM
Lar Pressure Discharges. Winkler and his co-workers2-5 have systematically
studied the reactions of nitrogen, activated by passage through a high-voltage
discharge, with various hydrocarbons. These studies were carried out at pressures
near 1 torr with very low reagent flow rates. Gas temperatures were also low,
typically 300 C. In this nonequilibrium system the reaction betw-een methane and
"active" nitrogen yielded hydrogen cyanide, acetylene and hydrogen2.
High Pressure Arcs. More recently high-power thermal arcs were used for the
study of the H-C-N system near atmospheric pressure. Leutrer' operated 2 nitrogen
plasma jet into which methane was mixed. In his briefly reported results, up to
LO percent conversion of the nitrogen to HCN was found'. -At this symposium
Freeman? has reported an extensive study of the synthesis of HCN, from EL nitrogen
plasma jet intermixed with methane. Typically, a 7 percent conversion of N2 to
HCN was observed. In both studies &, C2IJ2, and unreacted CHI, were also identified
as major constituents in the cooled product stream.
If the maximum mean temperature attainable in these previous studies is cal-
cdated, one finds that enthalpy input was insufficient to generate mean tempern-
f Caes greater than - 5000 K.
\
J
,'
B
d
P
Dissociating Plasmas. Upon considering the various species possible in.the
H-C-N system, the strongest bond found is N E N (226 kcal/g-mole). Thermal-dissociation
of N2 is essentiaU.y complete at temperatures in excess of 8000 K". Orv.
car. envision a high-pressure, stable plasma fed with any of e. variety of carbon,
hydrogen, or nitrogen compounds and supplied with sufficient erthalpy to reach
this high temperature range. The constituent species of srlch a plasma would be
prinarily atomic nitrogen, atornic hydrogen, and atomic carbon near lo1-( cm-3 concentration.
The result of rapidly quenching such! highly reactive atom mixture
nas been generally Lnexplored. &ann and TimminsJ,lO, however, did study the
rapid quenching of the simpler atonic-nitrogen/atormc-oxygen mixtures at 10,Oc)O K

348
and 1 atm, which were generated by a constricted d.c. arc. The study which is
reported in this paper was undertaken to explore the chemical composition of
mixtues formed. by the rapid . .quench . of atomic-nitragen/atomic-carbcn/atomichydrogen
rnixtures.
PLASMA REACTOR
Radio-frequency Induction Plasma. One convenient approach to achieve the
reqdred temperatures in excess of 10,000 K at atmospheric pressure is generation
of a thermal plas,m by radio-frequency induction heating. The techniques for
generating and containing a stable high-temperature induction plasma have been
described in detail by Reedljll, Mironeru, Marynowski and Monroe'3, Freeman and
ChaselL, and Thorpel?. Figure 1 presents a diagram of the plasma generator used,
A few turns of copper tubing were wound around the outside of a water-cooled
quartz reactor tube. This coil inductively coupled the supplied r.f. power into
the gaseous reactants sent through the reactor. A water-cooled sampling tube with
very small internal diameter which served to quench rapidly the reactive plasma
species is shown. Efficient cooling allowed the entrance tip of the quench tube
tc be placed directly into the plasma core.
A cross-sectional diagram of the reactor is presented in Fig. 2. The iso-
therms indicated are those determined spectroscopically by Reed' and' are repre-
sentative of conditions in this study. No direct temperatwe measurements were
attempted in the present study.
The Induction Plasma as a Chemical Reactor. Several characteristics of this
system can be exploited for chemical synthesis:
1. Average temperatures in excess of 10,000 K allar essentially complete molecular
dissociation. (Lower power input can reduce the specific enthalpy to
preserve desired free radicals.)
2.
Plasma stability is maintained while operating at very low gas velocities
(

I 1
349
c
M
m
;
a
e,
k
M
x

350
PLASMA COMFOSITION 1
Thermochemical Equilibrium. One of the unique features of the thermal induction
plasma is stable operation with low gas flow. Since flow velocities in the
plasma zone are typically a few centimeters per second, residence times on the
order of a second are associated with plasma species. In a plasma near atmospheric
pressure the particle mean free path is very short and the collislpn rate very high.
At plasma temperatures (c. 10,000 K) gas phase reactions can be expected to proceed
quite rapidly. These facts lead to the initial assumption that local plasma
composition approaches thermodynamic equilibrium.
1
,
\
The H-C-N System. The theoretical compositions of hydrogen-carbon-nitro en
mixtures at thermod namic equilibrium have been computed by Kroepelin, et al. f6 ,
Mrynowski, et al.'f, and Bronfin, et a1.l8, for a variety of conditions. Computations
were based on free energy minimization using tabulated thermochemical datalg.
Figure 3 presents calculated equilibrium cornposition data for a typical stoichiometry:
H:C:N = 4:1:2, (CH4/N2 = l), at 1/2 atm (380 torr). Only species whose concentration
is greater than 1.0 mole-percent are shown on this plot; however, twentyone
different chemical species were considered. In Fig. 3a, full equilibrium was
considered for a two-phase system which included graphite. Figure 3b shows that
temperature segment which is altered by the exclusion of the solid-phase, C(s).
Molecular nitrogen, N2, is a major constituent over a broad range up to 8000 K;
thereafter thermal dissociation results in the predominance of atomic nitrogen, N.
As mentioned above, at temperatures over 8000 K, not shown of the graph, the system
becomes completely dissociated into a simple three-component atomic state: H, C, N.
At temperatures greater than 7000 K significant thermal ionization occurs, gener-
ating significant concentrations of singly-ionized atoms, e.g., C , H , N'. As
noted in both plots, CHq dissociation is well underway at 1000 K, resulting in the
formation of I+ and C(s) in the two-phase system (a); but the formation of HCN and
C2& in the single-phase system (b). At temperatures above 3000 K, Q% begins to
fragment to QH and H, and with increasing temperature, to the atomic species. At
temperatures above 3800 K, HCN begins to fragment to CN and H, and with increasing
temperature, to the atomic species. Nitrogen-hydrogen species, e.g., NIQ, .occur
at concentrations below 0.01 mole-percent over the entire range plotted.
'\
i
'I
'
'\-
+ + I
In comparing the common temperature segments of the single-phase and two-phase
composition diagrams, important differences are noted in that: (1) higher concentrations
of species like HCN and C2& ar~ preserved at lower temperatures in the
single-phase case, and (2) the maximum concentration of these species is somewhat
higher in the single-phase case. Hence in the lower temperature range, the pre-
dlcted yleld of HCN and QIQ is enhanced by retarding carbon nucleation. This .; ,
'
kinetic limitation acts to freeze the mixture composition at around 2700 K so that
negligible composition change is predicted over a fairly broad temperature interval,
down to 1500 K.
Composition of the Plasma. No direct determination of the plasma composition
was attempted in the present study. Only a few efforts in this direction have 1
,
1
\
'
\
\
\
)
'
1
P
5
i
1
i

1.0
I
- a.
I
n
TEMPERATURE - OK
Figure 3 Equilibrium conposition of equimolar CH&-N* mixture. (a) Including solid carbon; (b) Excluding solid
carbor.. Region above T, common.
I
0
70
60
x)
40
30
20
IO
NITROGEN GONVCRSK)N
0 10-1 2 5 100 2 5 IO' 2 5
P41'IW -
-
Fig'xe 4 yieid cf HCI:, expressed as percent Pi2 converted, as a function of input stoichiometry. Solid lines
shov .win-- yield predicted by thermodynaniic equilibriwn. Lata points shov experimental results
frx. the plas-a reactcr: ~r flow rate 42 std cc/sec,. total reagent flow rate 2 std cc/sec, net
pmer - 3.5 kv. Reactor pressure identified with key. Points labeled x from plasma Jet studies6J7.

352
appeared in the literature. O'Halloran, et al?', have directly sampled an argon
plasma jet using a specially designed entrance cone opening to a time-of-flight
mass spectrometer. Raisen, et a1.21J have made spectroscopic identifications of
species in an air plasma. Unlike the mass spectrometer measurements, however,
emission spectroscopy 1s difficult to quantify due to the wide variance in the
oscillator-strengths of the likely emitters.
I
Faced with the difficulty of determining plasma composition directly, one is
prone to apply equilibrium predictions as 8 guide to local plasma composition.
Referring both to Figs. 2 and 3, a highly dissociated composition can be expected
in most of the plasm region. Temperatures in excess of 10,000 K, expected over
most of the central region of the plasma, would dictate that the atomic species
H, C, and N, and their ions would predominate.
QUENCHING
At this juncture it is important to ask what Changes in composition would be
encountered on cooling the €I-C-N atom plasma. A slow, gradual cooling of the
labile intermediates resident in the plasma zone may well allow the system to
revert to the original reactants along an equilibrium path.
Kroepelin and Kippid2
found tNs effect in their study of a 10 amp, 40 volt d.c. arc burning in 8 hydro-
carbon-nitrogen atmosphere. The composition of the slowly cooled arc-heated gases
was found to be mainly w, &, and C1 and C2 hydrocarbons. No HCN was observed.
Under extremely rapid cooling, however, kinetic limitations can interpose along
the reaction path to yield different, uwre interesting or valuable products.
The Cold-Wall Tube. A small-diameter, water-cooled tube was selected from
the variaty of available high-cooling-rate devices, to provide rapid and convenient
quenching of the plasm species. Figure lb shows a diagram of the simple design
of the three-concentric tube arrangement of the quenching probe23. The overall
outside diameter of the probe was 318 in. The internal diameter of the quenching
tube which receives the hot plasma was 0.032 in. The surfaces of the inner tube
were stainless steel; other parts were fabricated from copper. The effect of a
variation in the composition of the cold surface waa not studied.
Cooling Rate. Due to the Ngh rate of heat transfer from plasma to adjacent
cold wall, rapid cooling occurs in the quenching tube. Freeman and ~ltrivan~~,~5
have measured an initial temperature decay rate of 5 x 107 Vsec in water-cooled
tubes. In their model of the heat-transfer process occurring in very smaU tubes,
Anmann and Timminsm have calculated cooling rates greater than lo9 Vsec. *obably
quenching rates in this range were associated with the quenching tube used
in. this study.
Reaction Path During Quench. Unfortunately, there is a sparsity of hightemperature
kinetic data for the H-C-N system. Hence one is unable to predict with
certainty the reaction path followed as the atomic species H, C, and N, are cooled
from 15,000 K to 500 K in 10 11tI~iBeCOnd6, for example. After an examination of

353
the experimental results in the succeeding sections, it may be possible to infer
th important steps in the reaction sequence.
EXPERIMENTAL CONDITIONS
Radio-frequency SuppQ, A commercially 12 kw (n&mind) induction
heater, oscillating at 4 mHz, was the pmer source for the experiment. The load
coil, shown in Fig. 1, consisted of five turns of 1/4-in. o.d., water-cooled, copper
tubing wound tightly around the quartz reaction tube.
coil was 1 1/2 in., with a central diameter of 3 in.
The overall height of the
Reactor. Containment of the plasma was accomplished within a 35 cm long,
47 mm 1.d. quartz tube, mounted vertically. The high-power loadings necessitated
cooling which was afforded by flowing l/3 gpm of water into a cooling jacket surrounding
the central reaction tube. The plasma-forming gases were premixed and
sent into the tube through a brass fitting sealed onto the tube base. 0a6 flow
rates were monitored with calibrated rotameters, Plasma-heated gases left the
reaction tube through a second brass fitting sealed onto the top of the tube. This
cap, which WBB of approximately 1 liter volume, was water-cooled. In this configusation
the top fitting functioned as a cooling chamber to reduce the average
gas temperature to within a few degrees of ambient. The system pressure was controlled
by a high-capacity regulated vacuum line attached to the upper cap.
A sliding O-ring seal was provided at the center of the upper cap for positioning
of the 3/8-in. 0.d. by O.032-ln. i.db quench probe along the central axis
of the reaction tube. Cooling water was supplied to the quench probe at a metered
rate of l/3 gpm. The entrance tip of the probe WBB typically located at the center
of the uppermost winding of the load coil.
The flow rate through each cooling water line was metered with calibrated
rotamet.era wNch were placed downstream of liquid pressure regulators. Interconsistent
mercury thermometers were mounted at each cooling water line inlet and
outlet.
total enthalpy delivery rates could be determined.
& measurement of the temperature rise and flow rate in each cooling line,
Chemical Analysi.8. The gae stream withdrawn through the quench probe wae sent
to an on-line gas chromatograph for quantitative analysie. A triacetin column,
recommen ed by IsbeU27, followed by a molecular sieve column, recommended by
Purnel128, was ueed for reeolution of chromatogram peaks. This configuration
allowed the detection of the following compounds with a thermal conductivity cell:
HCN, I@, Ar, N2, CQJ and various higher hydrocarbons. A parallel flame-ionization
detector was useful for detecting low concentrations of wdrocarbons, e,g., Cq,
CzQ.
Experimental Variables. Total pressure in the plasma reactor was typically
380 torr; some data were also acquired in a range from 160 to 760 torr. For the
data reported here the argon feed rate was 42 std cm3/sec (6.5 g-mole/hr).
This

354
flow rate insured good plasma stability for the available r.f. power and remained
with the capacity of the subatmospheric pressure regulating equipment. Added
reagent gases were fed at 1150th to l/lOth that rate, with the molal ratio varied
over a broad range: 0.1 5 CH4/N2 5 25. The power coupled into the gas mixtures
was -3.5 kw maximum, as determined from the summation of cooling water heating
rates.
I
In the later stages of the experiment various nitrogen-substituted hydrocarbon
liquids were fed to the argon plasma, namely: CH3CN, CH3C&CN, and CQCHCN. Special
modifications to the gas-flow system were made to insure injection of these
nltriles into the plasma. A part of the argon feed stream was split off to a
sealed gas-liquid bubbler containing warmed reagent. Since these compounds are all
relatively volatile , a substantial mount of nitrile entrainment occurred. Rather
than introduce the nitrile-laden argon stream into the relatively cool gas region
at the base of the reaction tube, a special injection probe was provided to introduce
the stream into the hot plasma region. This second water-cooled probe was
positioned through an O-ring seal in the base cap. The probe design was identical
to that previously described for quenching (cf. fig. lb). The tip of the injector
probe was placed at the center of the lareramst winding of the load coil.
rial exiting the injector thus was assured of entering the plasma zone.
kte-
EXPERIMENTAI, RESULTS
Plasma Stabilitx. Over the entire range of experimental. variables the plasma
was stable and brightly luminous. The luminous region of the plasma appeared to
fill about 90 percent of the cross-sectional area of the cooled quartz tube and
extended from the bottom to about 3 in. above the r.f. load coil. With high hydrocarbon
fluu rates a gradual build-up of soot occurred on the quartz tube walls.
Methane-Nitrogen Reactions
Products of Reaction. A quantitative analysis of the gas stream withdrawn
through the quenching probe was made for each run. Over the entire range of
stolchlometric studies, the methane fed to the plasma wea completely reacted; no
methane was detected in the 'quenched .gas stream. The following major reaction
products were identified: HCN, C2&, 9. Also present were Ar and unreacted 5.
Nitrogen Conversion. Since essentially complete conversion of methane was
observed for each run, the varying conversion of nitrogen was selected 88 an important
datum. If the mole fraction of a species i found in the quenched product
atream is defined ea xi , then the conversion of nitrogen, defined BB a is given
bY :

,)
355
The ceaswed conversion cf ni;rsgc.n xa-ied from 3 psrcent at the lowest
stoichionetry cocsidered, CFU.+/Id2 = j.1, 'io I niaximum of 70 percent at a methanerick,
stc' -b.~lo:.-.ctr;, -;.. CH4/32 = 2j. Phis higii vaiuc of iiitrogen fixation is considerablj
1:: excess of prs-,.ioGsly reported nitrogen conversion levels observed in
therclai piasna reactions. Prcsswe variaiions had no significant effect.
Qpi?al Product Conpcsition. Tabie I, below, presents the composition of the
prcd..ct stream produced under typicel conditions.
Teble I
Typical Product Composition
Heactor Pressure - 3S0 torr Net Power Input - 3.5 kw
Feed Rate Feed Composition
InpLt (std cc/sec) (mole-percent)
97.6
1.2
1.2
-
Total 42.8 100.0
Composition
(mole-percent 1 (mole-percent, &-free)
90.5
2.0a
1.0
0.3
0.2
trace
-
-
57
29
9
5
-
-
190 .0 100
:;cte e: Ccnpositioiis dcterzi:ied fri:n gas chrcmatogra:r,i; - + 10 percent.
As wzs seen in Fig. h, Table I shcws that about l2 percent of :!,e nitrogel1 fed at
tbls stoici:icnetry vas corlverted to hydrogen cyanide. Ilinety-nine percent of the
,?!ethane fed was convertei :o the gaseoiis prccl.icts hydrogen and acetylene and also

..
to an unmeasured small qusntity of solid product which deposited on the wall.
The minor disperity between the nitrogen and carbon material balances indicates
the small fraction of material lost from the reaction zone to deposition on the
reactor walls.
Nitrile Reactions
Products of Reaction. Sdl quantities of nitriles could be continuously fed
to the argon plasma by using the modified injection arrangement (described in the
previous section).
cooled probe was analyzed with no change in the gas chromatograph. Essentially
Mixed plasma which was quenched by withdrawal through the water- ,
complete conversion of the lower molecular weight compounds , acetonitrile , CH3CN,
and acrylonitrile, C&CHCN, was observed. A minor amount (-10 percent) of propionitrile,
CH3C+CN, was observed in the quenched gas stream, perhaps due to a
bypassing of the plasma core. Each of the nitriles produced the same reaction
products, a repeat of those produced in the methape-nitrogen study. High concentrations
of HCN, Q+, I@, and N;! were observed dong with the argon diluent. A
't
'
complete quantitative chemical analysis of these nitrile runs has not, as yet, been
performed.
1
DISCUSSION OF RESULTS
Variance in the Conversion Percentage. Considerable variance in the f'raction
of nitrogen converted was observed among repetitions of runs with constant stoichiometry,
net power input, and pressure. The data points plotted in Fig. 4 include
the maximum observed conversion levels for each stoichiometry. Sources of error
in the experimental procedure which could have scattered the data darn from maximum
conversion were :
1.
2.
3.
Improper adjustment of r.f. supply controls so 88 to mismatch the supply and t
plasma impedance, perhaps generating plasma irregularities
Minor air leaks which would add additional nitrogen to the plaema to altec
the stoichiometry from preset values
Deposition of carbon or hydrocarbon material on the reactor walls or, the
reverse, significant vaporization of carbon from the walls, to alter the pre-$ ,
set stoichiometry.
I
i
Thermochemical Equilibrium Predictions. As described in the "Plasma Composi-
tion" section, above, a full range of composition calculations were made assuming
thermochemical equilibrium. These calculations predicted that the plasma composi- '
tion within the r.f. coil- would be dominated by atomic species for the specific
power input used in this study. Perhaps the following additional consideration Of
these theoretical predictions can elucidate the processes occurring during rapid ~
quench. 'I
i
',
I
\
I
\
\
l,
<
,
\
\, i
I
i
\
i c

1
6
357
Freezing. 3jpotnesise thzf eqxi1ii.ri.m is followed du.ri!ig the initial stages
I'\
cf coccling withir, the quench t.&e. For he~xistic pill'poses,, assaw that equilib-
I ,.i.... - k.. IS cbeyed dwing te:nperat:ce decay tc\ epprosicately 3300 K, or more specifi-
I
-,. -G~L;, tc that temperat-re vh5r-e ?qKilibri,x:i predicts the maximum number of moles
cf FCII tc be famed.
P
Then assux tiiat eli reactions involving nitrogen species
, are froze:: at that point, sc chat even as ccoling continues negligible decomposi-
1 tio!? I;f the occurs.
i Chemical Reaction Kicetics. In :w,kiiig this series of seemingly unwarranted
ass.mptions, sone specific reacCiori retes have been implied to be very fast, while
, .otiisrs, very slow. Focuskg C.!I nitrzgen-containing species, the fcllowing types
cf reactio:. nave been ass;;risd tc be rapid iinaer the conditions developed in the '
1 ..
Q',ik:.x:iLng tube:
I
\
Y
i
1
i
14' + e- M
Further, the following have been assumed to be very slow:
HCD; - many steps - €I! + N2 + C (5)
To date a full description of the reaction kinetics of the H-C-N system, even
at moderate temperatures, is unavailable. In the recent shock tube study of
hhrshall, Jeffers, and Bauer29, preliminary results indicate that the equilibrium
in Reaction 4 may be achieved rapidly at elevated temperatures. However, the
evidence points to a rather complex reactior. mechanism for the thermal dissociation
of HCN, wherein many more steps are involved than mentioned here. The cornplexity
of this system was encodntererl in the earlier study of Robertson and
Pease3', and iil similar systems explored by Goy, Shaw and Pritchard3l.
Rapid dissociation/recombination rates have been determined for nitrogen,
ReactiOn 3, by Wray3*. Reactisn rate data is not available for Reacclon 2, nor
for otker reactions likely to be quite important in the H-C-8 system.
Atomic-ion/electror. recoxbination reactions for each of the three atoms in
the plasma are likely to be rapid33 relative to the quenching time.
In surnniary, existing reaction rate data is fa- from sicfficient to check the
assu-,pticns made abcut the reaction niectanisms apprcpriate to this H-C-N system.
kdiiti onal kinetic studies which generate presently unl.own reaction rates are
needed before an accurate reaction path cari be predicted theoretically for the
reacting H-C-N system.
I

358
Ccxposition at the Assumed Freezing Point. The validity of this freezing
approacn can be found comparing the predicted composition at the freezing temperature
witc the sbserved composition of tne quznched gas stream. The follcwing
variaoles are defined for Table 11, whlcn facilitates the comparison.
For the equilibrium calculatians let the input stoichiometry be charac-
terized by the molar ratio CHq] /[N2 , set equal to + ; let nN21be the
number of moles of N;1 intro6uced; ani let n: be the number of moles OP
species i present at equilibrium 2t a pressure, P, and a temperature, T .
Then Tf is establisned for a given and P, as the temperature at which
equilibrium predicts the quantity nlCN/nN2 is a maximum.
Table I1 shows the mole fraction of the predominant species for a Tf value
found in an equimolar input stoichiometry. As mentioned in the "Plasma Composition"
section, above, solid-carbon formation may be retarded in this system. Therefore,
a second set of mole fraction data have been presented in Table I1 which
excludes the species C(s) from the calculation.
Table I1
Calculated Equilibrium Composition at the Temperature
where HCN Yield is Maximized
Pressure - 380 torr - 4 P [CHJ+]/ [Np] = 1.0
x2
*e
HC N
0.421
0.252
0.097
others 0.030
Including C( s)
Mole Fractions
0.543
0.283
0.109
0.038. 0.065
0.144 -
0.020
0.497
0.266
0.120
0.073
0.025
0.005
Note a: Adjusted to allow for the reactions: C2H + H - C2%, 2H - h.
0.522
0.274
0.124
0.080
-

359
\ Frozen Compositior: Calcillation v5rsus Expcri;:icntally. 0bscrvt.d Ccwposition.
IC, indeed, the oriSinal assuption that important reactiotis occurrirg in the gas
'
withdrawn from the plasma into the cold quenching tube are .frozen.in ths vicinity
1 1 ,of 3000 K is, valid, then agree'ment should be found betveeri the cc::!pssitions listed
in Tables I and 11. Ir. comparing the argon-fret coinposition listed in Table I with
PO~UIM 2 in Table 11, rciatively $cod agrement is found.
This comparison has been extended to thc' ikll ranee of stsichiomctries studied.
i)n Fig. 4, the maximum nitrogen'conversion, amar-l , and am0,-2, has been plotted
' against input CH)+/N2-molc ratio, 4,. The s.dlscript 1 refers to calculations includ-
'
ing C(s); subscript 2, to excludine, C(s). Agrceinunt between the freezing mcdel and
crbserved results is very good over more than a two-decade range in stoichiometry
variation. These results strongly support .the freezing model in the overall reac-
;
tion path. With the existing scatter of the data, a delineation between the
single-phase and two-phase equilibrium predictions of composition at Tf is not
possible. Additional sxperimental runs may lessen this ambiguity. The retarda-
/ tion of carbon solid fcrmation, however, accoiints fur Less than a 10 percent change
in predicted maximum nitrogen conversion, an the average.
Freezing Temperature. The data do not allow an t.xact determination of a frcczing
temperature appropriate to the quenching process. keferring back to Fig. 3a, a
change of a few hundred degrees in the assumed Tf could account for the data points
which indicate 10 to 20 percent less than the two-phase equilibrium predictions of
maximum nitrogen conversion. Fig. 3b shows the HCN concentration plateaus to be
quite broad. Hence the nitrogen conversion prediction would be relatively insensitive
to variation in freezing temperature over the range from 1500 to 3000 K.
Future experimental refinenents may allow a mvre exact determination of Tf .
Observed Compositions in Nitrile Experiments. An analysis of the maximum
nitrugen convcrsion predicted by tiiernochenical equilibrium has not been nlade fi.r
the different nitrile inputs desckibed in tne "Expcrirnental Conditions'' zectiuri.
The similarity cjf .product ciistri!iiition and uf the rxtio of. HCN to !Q in the prcduct
poitito to the same reaction mechanisms as iii tile CiU+/l\I2 studics.
Correlatiuri with Studies by Other Investigators
If the proposed reaction mechanism is correct, then the yields found in other
studies of the synthesis of HCN by thermal reactions of hydrocarbons with riitrogeri
should not exceed the predicted values of a . Some results of the production of
HCN in plasma jets have been reported, as was mentioned in the "fieviolis Studies"
section. While the differences in design between the plasma jet and induction
plasma reactors are not detailed here, sample results from these other studies are
plotted on Fig. 4. The maximum nitrogen conversion vdues observed 'in the nitrogen
plasma jet experiments of Leutnerll (designated XL) and FreemanE (designated xF )
appear. These experimental data points are seen to lie below the maximum nitrogen
conversion prediction.
I

360
The phenomena ir. t:te pias::.& ::et experiments which convert less nitrogen than
the ,;axinurn pred3ct.Yd 6). the ?'re:czilig rnuuel my well be l.ir,ked to a [nixing or diff'L;sian
rate (CHq a:. i 32 were !~dt prcfliixta) or' tc temperatures 4nsur'ficient to reach
Thi;.qsesticn is ~ i ~ sub,).:et l : 12 a separ-te theoretical analysis now in progress
Tf .
by the author .
THE REACTION SEQUENCE
The experimental data support the fallowing description of the overall reac-
tion seqdence:
i. Plasma Reactic;;w. ' Feed reagerits. are dissociated to their atomic constitu-
ents in a therim, plasma. Some Tthermai ioiiizatiQn accurs at sufficiently
high ternperatu.-s.
CF& i 92 - c + H + N + C+ + H+ + N+ + e-
2. Initial Coohnt Plasma taken inco a small-diameter cold tube rapidly
cools witn chemical reactio:is following equlibriwn.
C + H + N + ions - H + CIV + C2H + N2 - HCN + @ + N2 + C2w
3. Frozen Reaction Emetics. Rapid cooling continues, but at a rate much
greater than the progress of the apparently complex reaction sequence
necessary to des%roy trle species HCN, C2H2, and %. Hence the composi-
tion of the cooiliig gas is frozen at the end of Step.2.
Evidence fyr the Freezing llbdel in Other Reacting Systems. The freezing model ,
likely is applicable to a wide variety of reacting thermal plasma systems which '
emplcy a rapid quench. hmnann and. Timiinsl5 found a mechanism involving the freezin&
of equilibratiag reactiuns 'ta be applicable to their study of the quenching of \
nitrogen-oxygetl plasmc. Ahed by a wealth of published rate data on chemical reactiotis
in air, they were ~ble to model thc time-temperature-composition history of
M-0 plasma cooling within mail-diameter tubes. A quite similar freezing tempera- 1
ture, Tf , was found at 35Oi;'K 1'c.r that system. The equilibrium composition of nitric
uxi?e, NO, at Tf was preserved during continued rapid coolfng.
\
CONCLUSIONS ' . , . . \
Mixtures of mechane and nitrogen can be fed continuously to a thermal argon
indluctiori plasma maintaified abcve 13,300 K. .A rapid quer.ch of the heated plasma
kxture produces HCN, C2@, anc &.
the nitrogen converted, range between 3 and 70 percent, a function of the input
,
Final yields of HCN, expressed'as a fraction Of I
stoicbiometry . i
1
I
'';
<
ti
c

I
1
\
361
-.
ne reaction proceecis 1.; t:ie :cap;ccc intion on of the feed reactants in
-rj ..._ pias:% to yield a mixt,ae cf 3, C, ani 3. As this nixt.xe is quenched, the
rsectia:: initially procee5s 2;cr.g ar. ey;llibri,.it? pth, for1:;ir.g HCN, CZiI2, H2: and
*.4,-,. EC~GW the tenperat33y whey- cli? equ:-ibriL;:. :field OS 'TCN is maximized, the
;?owcr r-;lrra:;gclrient reacsions etlci carto:. ::x ieatiori are f'rozen by the rapid cool-
;:.&, s- tha: negligible citeraticn in the des cc!:ipositic!-. occurs upon further
c 15.
I
i
11.
12.
/ 13.
P. R. hm&nn and R. S. Timnlns, A.1.Ch.E. J., 17, 956 (1966).
T. B. Reed, J. Appl. Phys., &, 3146 (1963).
A. Mironer, A.I.A.A. J., L, 2638 (i963).
C. W. IWynuwski and A. G. Mc mJe, "R-f Generation of Thermal Plasmas," in
Sigh Teaperatue Technolcgy ( .

14.
15.
Lc .
17.
18.
19 *
20.
21.
22.
23
24.
25.
26.
27.
362
y. P. Freeman and J. D. Chase, "Energ- Transfer Plechanism of and Typical
Gperarink Cnarazteristics for thi. T!it.riial Radio-Frequency Piasma Generator, "
Paper No. 216, 133th Meeting, TiIc EJectrxhaniical Society, Oct. 12, 1966,
Philadelphia, Pa.
M. L. Thorpe, Research/Devel~~:;ient, 2 (L), 28 (Jan. 1966). ,
I
H. Kroepelin, D. E. Kipping, and H. Pietruck, "Equilibrium Partial Pressures
in the Systems C + H2 + N2 and C + Ii;! .+ iJH3 ar Temperatures from io00 to
I2000 K,," in J. F. Masi and D. H. Tsai, eds., Progress in International
Research on .Thermodynamic anii 1'rni:sport Properties, Proceedings of 2nd Symposiu?
on Thermcphysical Properties , Jan. 24-26, 1962, Princeton, N. J.
(Amer. SOC. Mech. Eng./Acade:nic PI'css, N. Y., 1962), pp. 626-48.
C. W. Marynowski, et al., Ind. El)&. Chen. Fund., 1, 52 (1962).
B. R. Bronfin, et al., "Therrricchemical Equilibrium in the Carbon-IIydrogen-
Nitr.ogen System at Very High Temperatures, " 15th Chemical Institute of
Canada Chemical Engineering Conference, Oct. 25-27, 1965, Universitf Laval,
Quebec.
JANAF Tables of Thermochemical Data, Dow Chemical Co., Midland, Mich., Dec. 31,
1964.
G. J. O'Halloron, et al., "Determination of Chemical Species prevalent in a
Plasna Jet, " Report No. ASD-Tm-62-644, Bendix Corp. Research Leboratories,
Southfield, fich., 1964.
E. Reisen, et al., Appl. Spectr., 2, 41 (1965).
H. Kroepelin and D. E. Kipping, "Teniperature Distribution in an Electric Arc
Operated ir! B Iiydrocar.bon-Nit1.0221) Atmosphere," in J. F. Masi and D. H. Tsai,
e&., Prorress in International Research on Thermodynamic and Transport Prop-
-, erties Proceedings of 2nd Symposium on Thermophysical Properties, Jan. 24-26,
1962, Princeton, N.
pp. 649-55.
J. (Amer. SOC. Mech. Eng./Academic Press, N. Y., 1962),
M. N. Piooster and T. B. Reed, J. Chem. Pliys., s, 66 (1959).
M. P. Freenaq and J. F. Skivan, A.1.Ch.E. J., 8, 450 (1962).
J. F. Skrivan and W. Von Jaskowsky-, Ind. Eng. Chem. Proc. Design Devel., I,
371 (1965)-
Lepel High Frequency Labs, Inc., New York, N. Y., Model T-5-3.
R. E. Isbeil, Anal. Chem., 2, 255 (1963).

\ 36 3
'
h,
I
26. H. Purnell, Gas Chronatogaphy (John Wiley, N. Y., 1962) pp. 371-3.
29. D. ;&shall, P. Jeffers, S. H. Ba,Jer, Chenistry Dept., Cornell Univ., Ithaca,
M. Y., personal comnvAnicaticn, NOV., 1966.
33. 1:. c. Rcbertson and R. N. Pease,.J. Amer. Chem. Soc., &, 1880 (1942).
31. C. A. Goy, D. H. Shaw, arid H. 0. Pritchard, J. Phys. Chein., 9, 1504 (1965).
, 32. K. L. Wray, Res. Rept. 104, Avco Everett Res. 'Lab., Everett, Mass. (June,
1961).
J
,
33. D. R. Bates, A. E. Kingstcn, and'R. W. P. McWhirter, hoc. Roy. Soc. A., 3,
, 297 (1962).

THE CONVERSION OF i.iEEIHAiW IN AN ARC REACTOR
P. R. Anunann*
R. S. Timmons
V. Krukonis
The conversion of hydrocarbons into other products, namely acetylene,
has been of considerable commercial and academic interest during the past
fzw decades. Several types of arc devices have been reported for carrying
out ?;he pyrolysis of methane.
Thermodynanic calculations show that if the acetylene is to be re-
covered r'rori an equilibrium chemical mixture, high temperatures in the
range of 2000 to 35000K are required, depending on the carbon-hydrogen
ratio and the pressure of the system. Furthermore, kinetic considerations
indicate that the pyrolysis of methane is fast relative to the decomposi-
tion of acetylene, and that there is an optimum reaction time to maximum
both the conversion of the methane and the yield of acetylene.
The most desireable accomplishment of high temperature methane pyroly-
sis would be to effect the endothermic reaction.
2CH4 = C2H2 + 3 H2
High temperature chemical equilibria for the carbon-hydrogen system have been
computed by Blanchetl and Bauer and Duff2. Assuming that chemical equilibrium
is achieved in an electric arc reactor, the data indicate that the temperature
at which the acetylene concentration is a maximum. The kinetics of methane
pyrolysis and acetylene decomposition during heating and cooling ultimately
determine the yield of acetylene.
Anderson and Case3 developed a kinetic model to describe the reactions
between arc heated hydrogen and methane. ' Calculations provided information
on conditions, such as He/CHI+ ratio, and reaction time, to achieve optimum
conversion of methane to acetylene. The analyses indicated that conversion
is not sensitive to contact time, within 0.1 and 1.3 milliseconds, and that
speed of mixing is the most important parameter.
Reactions of methane and hydrogen with carbon vapor produced by an electric
arc were reported by Baddour and Blanchetl. In their experiments, high
acetylene concentrations of 20 to 40 percent in the product gas were found.
In an arc process, the reactions which determine the success of the process
occur during the heating step and also during the cooling step. Heating
of the reactant may proceed either as it passes through the discharge, or
when it mixed with arc-heated gases. Cooling of the high temperature gases
is usually carried out quickly, either by contact with cold surfacestop mix-
ing with cold fluids.
This paper describes an experimental study on the con-
version of methane which either (1) passed through a confined arc discharge,
or (2) was mixed with an arc-heated gas. In this investigation, an attempt
was made to understand the factors which affect the extent of methane pyro-
lysis ana the yield of acetylene.
-c Avco Space Systems Division, 201 Lowell Street, Wilmington, Mass. 01887

In experiments on the direct conversion of methane, the reactant was
psssed through a confined arc discharge and quenched by mixing hydrogen or
helium at several different quench to methane flow ratios. The net enthalpy
Of the gas from the arc discharge varied between 78 and 150 kca$ per gram
mole which correspond to equilibrium temperatures of 2500 to 33000K. The
conversion of methane, that was observed, varied from 65 percent at the low
enthalpy to 85-100 percent at the high enthalpy. With a helium quench, the
methane conversion varied between 90 and 98 percent, but at low He/CH4 ratios,
a large fraction of the methane went to carbon black, due principally to lower
quenching rates and greater acetylene decomposition. Methane conversion
was slightly lower with a hydrogen quench and the latter probably affected
the kinetics of methane decomposition.
The conversion of methane by mixing with arc-heated hydrogen was studied
in a plug-flow type reactor. Hydrogen was passed through a confined arc discharge.
Islethane was mixed with the hydrogen leaving the arc zone, and heated
and reacted as the gases flowed co-currently through a cooled pipe., The conversion
of the methane was correlated with the enthalpy of the hydrogen arc
gas. Conversions were low (15-20 percent) at low enthlpies, and were nearly
complete at high enthalpies (go-100 percent). These energies corresponded to
50-60 kcal per mole of methane at the low enthalpies to 220 kcal at high values,
and were very Large for the degree of methane reaction. Two effects
were important. The arc gas lost energy to the cool reactor walls and the
rate of energy transfer decreased with time due to a decrease in the energy
driving force. By allowing for energy losses, the conversion was correlated
and fair agreement was obtained between the data from (1) cracking methane in
the arc discharge, (2) cracking methane in an arc-heated hydrogen stream, and
(3) the experbents of Anderson and Casel.
The yield of acetylene was also
observed to be a function of the available energy, and under sane conditions,
acetylene yields of go to 100 percent& theoretical were achieved.
References
1. Blanchet, J. L.,"Reactions of Carbon Vapor with Hydrogen and with Methane
in a High Intensity Arc, " Sc. D. Thesis, M.I.T., June, 1963.
2.
Bauer, S. H., and Duff, R. E., The Equilibrium Composition of the C/H
System At Elevated Texperatures, LA-2556, Los Alamos Scientific Labora-
tory, Los Alamos, N. M., September 18, 1961.
3. Anderson, J. E., and L. K. Case, An Analytical Approach to Plasma Torch
Chemistry, 1, 3, 161-165 (1962).

_r
333
THERMAL CRACKING OF LOW-TEMPERATURE
LIGNITE PITCH
John S. Berber, Richard L. Rice, and Delmar R. Fortney '
U. S. Department of the Interior, Bureau of Mines,
Morgantown Coal Research Center, Morgantown, West Virginia
INTRODUCTION
The tar used in this study was produced by the Texas Power & Light Company
from a Texas lignite carbonized at about 500" C in a fluidized bed. The
pitch, as used in this research program, is the tar distillation residue boiling
above 350" C. This pitch is about 40 to 50 percent of the crude tar. Pitch is a
complex resinous mass of polymerized and polycondcnsed compounds (1). It is
an amorphous solid material and quite brittle at room temperature. Tce pitch
analyses average 84 percent carbon, 8 percent hydrogen, 5 percent oxygen, and
1 percent each oi nitrogen and sulfur. It is chemically similar to the tar from
which it is prepared, .being mainly mixtures of the higher homologs of the compounds
contained in the distillable fractions of the tar.
The pitch is potentially useful as a binder in the manufacture of products
such as roofing cement, metallurgical electrodes, asphalt paving, and pitch fibre
pipe, if its characteristics can be modified by physical and chemical techniques to
approximate those ol. asphalt and bituminous binders (2).
Researchers of the Bureau of Mines, U. S. Department of the Inter.ior, have
investigated three methods for changing the lignite pitch characteristics.
Air-blowing, which is used successfully to treat bituminous pitch by lower-
ing the hydrogen content and increasing the softening point and the penetrability,
was not too effective with lignite pitch (1, 2).
--
Catalytic dehydrogenation (a) was found to be effective in reducing the
hydrogen content of the pitch, but catalyst cost per pound of treated pitch is high,
and the catalyst recovery cost would be prohibitive.
Thermal cracking has proven to be the most effective in changing the pitch
characteristics. This' report covers the preliminary work done on thermal cracking
of pitch and shows the variety of products that may be obtained.
Thermal cracking is widely used in the petroleum industry, particularly for
the production of olefins. Thermal cracking has also been used in the treatment
of coal tar.(A, ?), but seldom has been used in the treatment of pitch. The successful
treatment of pitch by this process could upgrade the pitch into metallurgical
coke, binders, oxidation feedstock for phthalic and maleic anhydrides, and gases
such as hydrogen, methane, and ethylene.
The binding ability of the cracked pitch and oil distillation.pitch can be
determined by the quality of metallurgical electrodes produced using the pitches
as binders. Many factors affect the binding properties' of the materials being

367
used as blnders, with different producers and consumers using entirely different
standards, so the electrode quality tests were used to help evaluate the effective-
ness of thermal cracking of the lignite pitch.
EQUIPMENT
The thermal cracking system is shown in figures 1 and 2. The feed tank
(figure 2, A) is made from a 13-in. length of 10-in. diameter schedule 40 pipe,
to which a cone has been added for the bottom. The tank is heated by commercial
electric heaters. The feed pump (B) is a small gear pump with a variable speed
drive. The thermocracker (D) is a 4-1/2 ft-length of 2-1/2-in. diameter schedule
40, type 304, stainless-steel pipe, heated by external electric heaters. The
heaters are controlled by outside surface thermocouples welded to the reactor.
The receiver (E) is a 20-in. length of 4-in. diameter schedule 40 pipe. The
receiver is flanged so that it can be bolted to the bottom of the reactor. The condenser
(F') is a 30-in. length of 6-in. diameter pipe, swaged at the bottom end to
I-in. diameter and flanged at the top. The top flange supports a water-cooled
tube which inserts into the condenser body in the manner of a cold finger. The
knockout (G) is a 34-in. length of a 2-in. diameter pipe having a tangential gas
inlet about 8-in. from the bottom. The scrubber (H) is a 30-in. length of a 4-in.
diameter pipe having a water-spray nozzle near the top. The water from the
scrubber drains into a standpipe water-seal tank (I). The water-seal tank is a
piece of 6-in. dlameter pipe. The gas meter (J) is a laboratory-type wet-test
meter capable of metering 1,000 cu ft of gas at standard conditions.
PR OC E DURE
Lignite pitch at 200" C was pumped from the feed tank through electrically
heated lines into the top of the thermocracker. The temperature of the thermocracker
was maintained at approximately 790" C as indicated by the outside
surface thermocouples. At the end of a run, the flow of pitch was stopped and the
pump was' flushed with a low-boiling tar fraction to keep the impeller of the pump
from freezing. Reactions within the cracker produced coke, cracked pitch, oil,
and gas. Both the top and bottom of the cracker were removed, and coke was
removed from the sides and weighed. Cracked pitch caught in the receiver was
weighed and analyzed. The oil removed by the condenser and knockout chamber
was weighed and then distilled to 400" C, leaving a pitch residue. This pitch
residue was analyzed and, if found to have the desired carbon-to-hydrogen atomic
ratio (1. 20 to 1.80), was used as a binder for electrodes. Distillate from the oil
may be oxidized to phthalic and maleic anhydrides or separated into acids, bases,
and neutral oils. Gas from the cracker was cooled and scrubbed, then metered,
sampled, and analyzed by gas chromatography.
RESULTS AND DISCUSSION
Results of the thermal cracking tests are given in tables 1 to 4. Of the four
major products obtained from the pitch--coke, cracked pitch, oil, and gas--the
coke averaged about 15 to 25 percent of the feed to the cracker, the cracked pitch
amounted to about 20 to 40 percent, the oil was about 20 to 30 percent, while the
balance of the feed to the cracker consisted of gas.

Distribution and yield rates of products (figures 3 to 6) are as expected considering
the reaction within the cracker at different crude,pitch.feed rates. As
the pitch is heated it begins to vaporize, and the turbulenc'e of the vapors splashes
some of the iluid onto the hot wall where it sticks and is coked. If enough heat is
available, the rest of the pitch is vaporized and cracked. ' As the cracked
materiais leave the hot zone, heavy high-b

369
'
i
1.
REFERENCES
Berber, John S., and Richard L. Rice. Oxidation of a Low-Temperature
Lignite Tar Pitch. Preprints, Div. of Fuel Chemistry, Am. Chem. Soc.,
v. 8, No. 3, Aug. 31-Sept. 3, 1964, pp. 145-151.
'
,
2. Chelton, H. M., R. N. Traxler, and J. W. Romberg. Oxidized Asphalts in a
Vertical Pilot Plant. Ind. and Eng. Chem., v. 51, No. 11, dovember 1959,
pp. 1353-1354.
I
' 3. Greenhow, E. J., and Calina Sugowdz. The Chemical Composition of Coal-
I Tar Pitch. Coal Research in CSIRO, No. 15, November 1961, pp. 10-19.
' 4. Montgomery, R. S., D. L. Decker, and J. C. Mackey. Ethylene and
Aromatics by Carbonization of Lignite. hd. and Eng. Chem., v. 51, No. 10,
October 1959, pp. 1293-1296.
I'
1
I
i
J
I 'I
\
J
5. Naugle, B. W., C. Ortuglio, L. Mafrica, and D. E. Wolfson. Steam-
Fluidized Low- Temperature Carbonization of High Splint Bed Coal and Thermal
Cracking of the Tar Vapors in a Fluidized Bed. BuMines Rept. of Inv. 6625,
1965, 22 pp.
6. Rice, Richard L., and John S. Berber. Catalytic Dehydrogenation of Low-
Temperature Lignite Pitch. Presentation, Div. of Fuel Chemistry, Am. Chem.
Soc., March 22-31, 1966.

367
used as blnders, with different producers and consumers using entirely different
standards, so the electrode quality tests were used to help evaluate the effective-
ness of thermal cracking of the lignite pitch.
EQUIPMENT
The thermal cracking system is shown in figures 1 and 2. The feed tank
(figure 2, A) is made from a 13-in. length of 10-in. diameter schedule 40 pipe,
to which a cone has been added for the bottom. The tank is heated by commercial
electric heaters. The feed pump (B) is a small gear pump with a variable speed
drive. The thermocracker (D) is a 4-1/2 ft-length of 2-1/2-in. diameter schedule
40, type 304, stainless-steel pipe, heated by external electric heaters. The
heaters are controlled by outside surface thermocouples welded to the reactor.
The receiver (E) is a 20-in. length of 4-in. diameter schedule 40 pipe. The
receiver is flanged so that it can be bolted to the bottom of the reactor. The condenser
(F') is a 30-in. length of 6-in. diameter pipe, swaged at the bottom end to
I-in. diameter and flanged at the top. The top flange supports a water-cooled
tube which inserts into the condenser body in the manner of a cold finger. The
knockout (G) is a 34-in. length of a 2-in. diameter pipe having a tangential gas
inlet about 8-in. from the bottom. The scrubber (H) is a 30-in. length of a 4-in.
diameter pipe having a water-spray nozzle near the top. The water from the
scrubber drains into a standpipe water-seal tank (I). The water-seal tank is a
piece of 6-in. dlameter pipe. The gas meter (J) is a laboratory-type wet-test
meter capable of metering 1,000 cu ft of gas at standard conditions.
PR OC E DURE
Lignite pitch at 200" C was pumped from the feed tank through electrically
heated lines into the top of the thermocracker. The temperature of the thermocracker
was maintained at approximately 790" C as indicated by the outside
surface thermocouples. At the end of a run, the flow of pitch was stopped and the
pump was' flushed with a low-boiling tar fraction to keep the impeller of the pump
from freezing. Reactions within the cracker produced coke, cracked pitch, oil,
and gas. Both the top and bottom of the cracker were removed, and coke was
removed from the sides and weighed. Cracked pitch caught in the receiver was
weighed and analyzed. The oil removed by the condenser and knockout chamber
was weighed and then distilled to 400" C, leaving a pitch residue. This pitch
residue was analyzed and, if found to have the desired carbon-to-hydrogen atomic
ratio (1. 20 to 1.80), was used as a binder for electrodes. Distillate from the oil
may be oxidized to phthalic and maleic anhydrides or separated into acids, bases,
and neutral oils. Gas from the cracker was cooled and scrubbed, then metered,
sampled, and analyzed by gas chromatography.
RESULTS AND DISCUSSION
Results of the thermal cracking tests are given in tables 1 to 4. Of the four
major products obtained from the pitch--coke, cracked pitch, oil, and gas--the
coke averaged about 15 to 25 percent of the feed to the cracker, the cracked pitch
amounted to about 20 to 40 percent, the oil was about 20 to 30 percent, while the
balance of the feed to the cracker consisted of gas.

Distribution and yield rates of products (figures 3 to 6) are as expected considering
the reaction within the cracker at different crude,pitch.feed rates. As
the pitch is heated it begins to vaporize, and the turbulenc'e of the vapors splashes
some of the iluid onto the hot wall where it sticks and is coked. If enough heat is
available, the rest of the pitch is vaporized and cracked. ' As the cracked
materiais leave the hot zone, heavy high-b

369
'
i
1.
REFERENCES
Berber, John S., and Richard L. Rice. Oxidation of a Low-Temperature
Lignite Tar Pitch. Preprints, Div. of Fuel Chemistry, Am. Chem. Soc.,
v. 8, No. 3, Aug. 31-Sept. 3, 1964, pp. 145-151.
'
,
2. Chelton, H. M., R. N. Traxler, and J. W. Romberg. Oxidized Asphalts in a
Vertical Pilot Plant. Ind. and Eng. Chem., v. 51, No. 11, dovember 1959,
pp. 1353-1354.
I
' 3. Greenhow, E. J., and Calina Sugowdz. The Chemical Composition of Coal-
I Tar Pitch. Coal Research in CSIRO, No. 15, November 1961, pp. 10-19.
' 4. Montgomery, R. S., D. L. Decker, and J. C. Mackey. Ethylene and
Aromatics by Carbonization of Lignite. hd. and Eng. Chem., v. 51, No. 10,
October 1959, pp. 1293-1296.
I'
1
I
i
J
I 'I
\
J
5. Naugle, B. W., C. Ortuglio, L. Mafrica, and D. E. Wolfson. Steam-
Fluidized Low- Temperature Carbonization of High Splint Bed Coal and Thermal
Cracking of the Tar Vapors in a Fluidized Bed. BuMines Rept. of Inv. 6625,
1965, 22 pp.
6. Rice, Richard L., and John S. Berber. Catalytic Dehydrogenation of Low-
Temperature Lignite Pitch. Presentation, Div. of Fuel Chemistry, Am. Chem.
Soc., March 22-31, 1966.

. .
I
. I
370
. . . .
., . . ..
. .
'',..,' , :.,
. . ..
. .
i
1
't

I
7
/*
i
,I
{
\
J
7
1/
r-'
I
Ia Jd - I
PI -------
I
I

4
372
090-
> E w - m c
N O - 0
z : 0000
. . . .
L o
3

I'
p'
/
!
2'
9'
c
u
.r( Y
a
e,
.r( Y
c
e0
d .r(
e,
I4
* 1
m
0
8
Y
I
: d
u
0
M
c
.r(
5
m
u
d
g
e,
5
k
0
w
e,
2
d
4
n
4
m
.r(
I4
0)
Y
2
I
I
373
. . .
. . . . . . . .
0 - N N N N m m
I 1
I
. ,

A-Feed tank
6-Thermocracker
GReceiver
D-Condenser
374
FIGURE 1. - Pitch Thermal Cracking Apparatus.
. . A.F.ed lank
B.Pumo and drih
C.Purnp (lush tank
D.Thermocracker
J
my water
FIGURE 2. - Thermal Cracking System.
I 1
\
\
\
\
\
t
i
\
Y
1
Y
'r
c

FIGURE
375
I I I I I I I I I I I
0 4 8 12 16 20 24
FEED RATE, Ib/hr
3. - Coke Rate Based on Crude Pitch Feed
Rate - Thermal Cracking.
14
12
10
2
0 4 a 12 16 20 24
FEED RATE, Ib/hr
FIGURE 4. - Product Pitch Rate Based on Crude Pitch
Feed Rate - Thermal Cracking.
I

80
\
0
70
60
e 40
a U
VI
9 30
20
10
0 4 8 12 16 20 24
FEED RATE, Ib/hr
FIGURE 5. - Cas Production Rate Based on Crude Pitch
Feed Rate - The'rmal Cracking.
8
- e4
0a i k6: 2 -
I I 1 - 1 1 I I I I I '
_,J: -
-0 I I I I I I I I I I
i
-

I
,
1
1
I
4 9
I
1'
i
J
20
0
377
Llllilllrlrll
4 8 12 16 20 24
FEED RATE, Ib/hr
FIGURE 7. - Carbon-to-Hydrogen Atomic Ratio of Oil
Distillation Residue Based on Crude Pitch
Feed Rate.

COAL DEASH& AND HIGH-PURITY COKE
Ward J. Rloamer , The Lummus Company, Newark, Ne J, and
F, L, Shea, Great Lakes Rasearch Corp., Elizabethton, Tennereea
I--" tLC A O ~ Ut C 132
In 1957, The Lummus Company and Great Lakee Carbon
Corporation began a joint investigation into the qse of coal
as a source of high-purity coke. The process involved the pro-
duction of a low-ash, low-sulfur deaohed coal ao1,uaion from
high volatiie bituminous coal, and the conversion of this coal
solution into coke using a modification of the Lummus delayed
coking operation,
Experimental work on a bench scale was initiated in
1953 and completed in 1961. The experimental stope was expanded
to include the production of deashed coal as well ae the high-
purity coke in either 'combined or alternate operetianso
Bench programs were carried oat on several scales of
ogeration. A contifiuous (block operation) pilot plant was
built and operated to conf%rm the bench work. This unit was cap-
able of processing coal continuously at approximately 200 pounds
per hour through the filtration step. At this point, the filtrate
was stored for further continuous proceseing to either drashed
coal product (performed in 150-200 pound quantities) or high-
purity cake (500 pound quantities) .,
-
S unna rv
A process has been developed to upgrade the quality of
coal to a low-ash, low-sulfur, low-chloride product, As a fuel,
this material would result in reduced capital and operating coati
for power stations, Investment costs would be reduced, due to
the higher heating value and cleaner fuel. Operating costs would
be reduced by charges associated with the transportatfon of ash,
ash hacdling and disposal costs, lower maintenance coetr, and fly
ash removal costsc The substantial reduction in eulfur, ash, and
volatile matter would greatly minimize air pollution.
Other potential applications are for use in gas turbine#,
as a reductant, as a raw material for synthesis gas, aa an ingredien
ia preparing coal blends for metallurgical coke, or for high-purity
coke.
)'
The economic potential of the process in its present stage '
of development is covered in Table 5. The process has certain i
aspects requiring sustained demonstration efforts to obtain firm
econonic bases.
1.

Process Descrivtion
379
Figure 1 shows a schematic,flowsheet for the production of
either deashed coal or high-purity coke as investigated in this pro-
gram,
I
Deashed Coal - Crushed, ground and dried raw coal is fed to
agitated premix tanks where it is mixed with warm solvent. The
slurry is pumped through an extraction heater and thence through
either in-line reaction tubes or to agitated digestion tanks. Diger-
tion was investigated over the temperature range of 650 to 850'F and
over a pressure range of 75 to 135 psig. Residence times varied from
three minutes to two hours. Solvent-to-coal ratios were varied from
l r l to 4:l
From the digestion tanks, the coal dispersion 1s fed to a
rotary pressure precoat filter designed for operating at.100 paig
and 700°F. Washing provisions are included.
The filter zake (L)fr discharged into a cake receiver,
Filtrate and wash are collected and fed to the product and rolvent
recovery operations, Vapors from the filtrate receiver are cornprersed,
heated and recycled to maintain pressure on the feed side of the fil-
ter drum,
Filtrate flows through the filtrate heaters into a vacuum
product recovery unit, Vapors from this unit pass to a flash tower
where the bottom stream is all solvent for recycle, A side stream
is taken off and in part is used for solvent make-up with the re-
mainder processed for by-product recovery.
Overhead vapors from the flash tower are condensed, and a
low boiling fraction recovered as a by-product. Vapors flashing
from the filtrate receiver are condensed and introduced as reflux into
the bottom section of the flash tower,
Filter cake is calcined and the epent cake used as fuel for
the process, Gas and distillate vapors from the kiln combined with
gaseo from the filtrate receiver and flash tower overheads are passad
into a light oil scrubber. The absorption oil -- the intermediate8
cut from the flash tower -- is stripped in the flash tower, Light
oil fractions leave the rystem from a decanter on the flash tower
overhead stream. Gas from the light oil scrubber is ured for procesr
heat
High-Purity Coke - For coke product the same flow pattern
prevails through the filtrate receiver. The filtrate is pumped to"
a coking heater and thence to coke drums. Overhead vaporr from the
coke drums pars to a combination tower where recycle rolvent and by-
product fractions are recovered, Overhead vapors from the tower parr
I

3Pn
_. -. - L -J:,J.-;~~~L, ):as separator, and thence to a light oil scrubber.!
at~ave operations were investigated on both 8 bench ,
..:.- ;i;: ?L;,nt scale with the exception of cake crlcining which was
?cr:azn~2 oniy on bench scale, Variables investigated are discusred ,
I
... i.16 :oiAowing sections.
'racLDJ ,nd Oneratint! Variables for Deashed Coal
- I-
,I reviev of our work on solvent deashing indicated the
07crntiny. conditions necessary to optimize the process. These
Z L ~ be briefly summarized:
.~
i: .The development of a stable, high-extractive-efficiency
i~iviii: cripabla of essentially complete recovery in the vacuum recov-
2:y uniz 'ca yreld ueashed coal was accomplished by special treatment
sz L i~i:~::-b~iii~g coal tar distillate. The required solvent having
ascelia:-.: ccmperacure ~ t a blity i and high solvent power was attained
by rcfra-:otrLing the hifih-boiling coal tar distillate, employing
" -LL>c>e ,I. t:iarmal decomposition and fractionation to eliminate the leas
- -Li-4L-o--y , .- . -. and low-bailing constituents. Gratuitously, these are the
;CL= cffectivi ercraccion components,
2 Xicsve:y of from 88 to 95 weisht percent of the avail-
3bie ~arbsnac-s~s matter at extractor preesures in the range of 5 to 8
.i;-isJ?heiLs and ri;h solvent-to-coal ratios in the range of 2 or 311
was Ychi?ve:d using ;he iefractorized solvent with A boiling range of
600 ~a 900-F Typical extraction conditione were 750°F, 75 psig, at
a asLven;-io-co~i :de10 ot 311.
2, Extraction at residence times of 5 to 20 minutes in a
-.c-;:r3u s~i~tizer vessel orB preferably, in continurvs in-line
ssi~rroa n~~c:ng' coils which would permit a direct circuit from the
ao:vrct-csz. mixing tank through a conventional heater with a discharge
to dr~her a rotary pressure precoat filter or a suitable
tyCro:lond,% i3i Bench extractions were successfully conducted at
no1G:ng t.irnSs ;f as low as 3 minutes. There were duplicated in a
piic: f ~ ~ i i - hiater ~ e coil discharging the hot coal rolution continuously
~:hr~,gh cr, expanded section of the transfer line to the filter
charga tank,
4. Ash removal from the coal solution WAO accomplished at
i:-ir.i~ror, r.i~cs of 200 to 300 lbsfhrleq. ft, of filter surface,
?=el:cindry rates of 200 KO 250 were secured in the pilot rotary
.,;J;OL~ prcss~re filter at low pressure differentiala, Small-scale
:-zb an cocit: washing and calcining were performed to obtain closediycie
data an soibent and distillate distribution. It was anticipa~od
that the filter cake would be washed with a light oil, calcined
zni the spexit cake used as nteam fuel for the procars in A full-scale
Tlznt,
\

381
The bench and pilot plant operations in general Closely
approached the goal5 set at the start of the work for the deashed
coal produstiJn phase of the program. In Table I, there are pro-
vided examples of the data relative to the composition of the raw
coal, coal-to-solvent ratios, temperatures and pressures and analyses
of the coal extract solutions, Yields are shown on he basia of
both raw and ash-free coals. Typical coal deashing results are
shown in Table 2,
Avai lable Coal Charge Stocks
Extraction of coal with phenanthrene by earlier workers
related the degree of extraction to the rank of the Thus,
about 95 percent of the organic material in bituminous coal was
dispersed as compared to 27 percent for subbituminous coal and 23
percent for lignite Bench studies using a preprocessed coal tar
distillate solvent marched these extraction efficiencies for the
bituminous coal and lignite studied,
The coals extracted ranged in ash content from 5 to 20
weight percent and in sulfur values from 1 to over 4 percent. In
each case, it was possible to secureexcelknt removal of the ash
and a subsranrial portion of the sulfsr, chlorides and other con-
t aminan t s - Sulfate and pyritic sulfur were fully eliminated with the
ash and the organic sulfur forms were noticeably reduced., It was
also possible to reduce the volatile matter contents from original
values of near 40 percent to around 15 to 25 percent by distillation
or by solvent precipitation (5) of the deashed coal solution,
Improvement of Solvent for Coal Deashing
Ihe wide-range heavy anthracene oil fraction originally
used in the coal solution studies boiled between 600'F and 1050°F.
Bench delayed coking of the deashed coal solutions gave solvent
losses ranging as high as 30 to 60 volume percent through polymer-
ization of the heavy ends and decomposition of the less refractory
components of the solvent.
Accordingly, studies were initiated to produce a more
stable solvent, As a result, the efficiency of the coal solvent
was improved to a point where 85 to better than 90% solution is
achieved of the total extractable carbonaceous material of the coal
(l.e., excluding the nonsoluble fusain)
This preferred polvent, freed of easily polymerizable
materia1,is capable of essentially complete recovery from either
the cos! solution by vacuum stripping or from the delayed coking
operation at reduced pressure. .
F

S ~ l ~ o n~pg:ading r is the result of two important factors.
r l ~ s t , the 201 irrit holling range has been narrowod and, secondly,
the 5ol~ent has S E ~ ~ ,cabilised I
by removal of its less refractory
(readllv crached, ,,mponents.
qolvent improve 1cb stahilitv.
Continued use and recycle-of this
A hciff dlscuaslon of the variables in coal solution and
coking of its deasned ~olutlon wlll outline the reasons for the
improvemenrs -e:.ireri by the channes In solvent boiling range and
composition noted

383
ash content of the initial high volatile bituminous coal. Thus,
it has been repeatedly demonstrated in the bench and pilot solvent
deashing studies on coal samples with ash contents varying from 5
to 20 weight percent that final ash values of 0.1 to 0.3 could be
obtained in the treated coal extracts.
Lisewise , inorganic chlorides have been iargely eliminated
with the ash. Coals containing 0.25 and 0.15 weight percent chlorine
were reduced to values of 0.004 to 0.003, respectively. It is
assumed that coals of considerably higher inorganic chloride contents
could be reduced to similar low values. The harmful corrosive
effects of ash and chlorides on furnace operation are well known and
their effective reduction is obviously very desirable.
Sulfur Reduction
The degree of bituminous coal desulfurization by solvent
deashing has proven to be a function of the original ratio of
inorganic sulfate and pyritic sulfur to the organic forms. Original
total sulfur contents of 1 t o 4 percent are reduced to values of
0.4 to 1 weight percent in the extracts.
It is known that this ratio increases with coals of
increasing total sulfur content, i.e., in the lower grade coals of
correspondingly higher ash. The final sulfur of the deashed coal
therefore relates to the original organic sulfur which is only
partially removed in the process, whereas the inorganic sulfur is
removed substantially quantitatively.
Thus, in Table 1, complete elimination of the pyritic
and sulfate sulfurs has been achieved with an accompanying 26
percent reduction in organic sulfur, for an overall sulfur reduct-
ion of about 50 percent.
Delayed Coking
The original 1957 concept of the coal deashing process
was to prepare a coal solution which, after filtration for removal
of its ash, sulfur and other contaminants, would be charged to the
conventional Lummui delayed petroleum coking process. The coke
drum vapor would pass to the combination fractionation tower for
separation of gas and light and medium distillate from a tower
bottom cycle oil which would be re-used for fresh coal solution.
The bench and pilot coking of the deashed coal Solution
proved to be almost routine. This charge stock resembled in many
of its characteristics a low temperature carbonization pitch.
Lummus had previously in bench scale and pilot plant delayed coking
apparatus successfully coked both low and high temperature carbon-
ization pitches as well as Gilaonite and Athabasca tar sand pitches.
n

The range ~f temperatures and general operating characteristics
proved similar to those employed for the stocks noted above.
(
Reduced press~re is necessary for high solvent (600-900'F) recovery.
Thus, it is anticipated that full scale plant operation would
produce cokes with volatile matter contents approaching those
normally encountered wich pecroleum stocks. Such cokes would be of
\
low ash and satisiactorv sulfur contents and would be secured at
high yields. \
The product distribution on coal solution and on coal are
summarized in Table 3, based on a 2:1 solvent to coal ratio. It \
was found on examination of the coker total liquid products that
the extraction solvent had been recovered without significant chnage. '
The properties of the coker feed and product coke are listed in 1
Table 4.
F
An alternate mechod of producing high purity coke utilizes ,
the pure coal excract as the raw material. This may be pulverized
in a ball and ring or xing roller type m i l l , such as is used for
powdered coal burners. The milled pure coal is then used to form a
slurry with a refractory recycle stock, having a boiling range of
about 500-700'F, The boiling range of this stock will vary somewhat,
depending upon conditions in the heoter, the coke drum, and the ratio
\
of solids to liquid, but is selected so that sufficient liquid phase 1
remains in the heater coils so as to convey the milled pure coal s
without coke build-up in the heater.
A third method of pure coke production considered,
particularly where production of pure coke from coal would be
conducted in canjunction wich the operation of a refinery, involves
utilization of petroleum coker feedstock as the liquid in which the (
slurry of pure coal is formed. Where a low coke yield feedstock is
used, the coke produced would be predominantly from pure coal. \
However, che proportion might be varied, depending upon the type of
coker feedstock employed and the ratio of solids to liquid in the
slurry charged LO the delayed coker.
Alternate Processes for Coal Extract Recovery \
As an alternate to the recovery of coal extracts of
original or decreased volatile matter contents by distillation, as
noted in Tables 1 and 3, it was determined that these products could .
be recovered readily from solution by precipitation with hydrocarbon ,
solvents. Thus, a paraffinic solvent yielded the complete deashed
coal extract, whereas it was possible to recover a deashed coal with
a volatile matter content 60 weight percent below that of the original
coal with control of intermediate product values by manipulation of
the aromaticity of the solvent blends.
c
4
\
\
,
,

385
Referencee
1. Bloomer, W. J., and Martin. S. W.. U. S. Patent 3,109,803
(November 5, 1963).
2. Dicalite 40, General Refractories Company, used for bench and
pilot filtrations.
3. Patent pending,
4. Orchin, M., Golumbic, C., Anderson, J. E.,
U. S. Bur. Mines Bull. 505, 1951.
and Storch, 8. H.,
5. Patent pending.
I
. . .

. ..
386
9
* -
r.
Y
2
.'
d
a
f'
r.
*
0
..
mo
00
..
m
m N
90
..
m
0
00
A0
In
O H .. *o
..
MN
dln
NN
I ,
N
mm
rc
IC-
-I M
a m
'i
c'
J
h
.
u? b
-
r,
C
i:
0)
C
..
ON
om
0-6
..
m.m
tnm
rn'
..
Inm
IC-.
.or-
..
,.
, .c
'.O
1
\

I
.-
4 -C .-
.-
J
u
0
-- 0
N -
In
VI
m l -
NVI VI
0
d
. \*
.. .
. .
- L (
Nlin
- 0
I l l
I l l

.)
.L
m
388
I

.
UI
m
rs
m
389
. . . .
t u 0 - r -
tu CI 0 9.
N 9
4
$
0,
Y
s
c
C
0)
d
0
ln
cr(
N
00.
0
N O
y;d
N 2
0
CI
I

TABLE 4
390
DELAYED COKING OF DEASHED COAL SOLUTION
RESULTS OF COKER TEST RUN DC-13
Properties of Feed.(’)
Specific Gravity, 100/60*F
Softening Point. *F (R&B)
Benzene Insoluble. Wt.70
Properties of Coke
Wt. Percent of Feed
Density, lbs/ft3
By Weight of Volume 50.1
Occupied in Coke Drum.
Coke Drum Cokc Drv.7
so. 1 Kc?. 2
1.1926 l.!6!l
158
22.2
25.7
By Weight of Water 66.7
’ Displaced by Sample
Volatile Matter, Wt. % 9.17
Fixed Carbon, Wt. % 90.22
Ash, Wte% 0.61
Sulfur. Wt. % 0.37
Heating Value, BTU/lb. 14,520
(1) Weight percent coal extract
Benzene insoluble, weight p
. . .. . . . .
. .
. . .
-. .. . .
r
-26.2
ent- 2 I., 0
ion. 5
I?. 6
22.3
38.5
51.3
10.06
89.’28
0.66
0.38
I
;
1
1
\
\

9'
-7
.I
\
f-
P
,
/
,/-
391
TABLE5
rr-ovx'rr STUDY - DTASHED COAL PRODUCT
D:ar.t Cost 513,400,000 d 24 000s OOO
Cost/Ton //MM BTU CostrTon
\
C'MM BTU
Product Product
Raw b! ate ria; ( Cc L 1)
Lzkor, SILT) e r ~ s : ion,
56.50
:. 14
21.6
3.8
$6.50
'
. .9R
, 21.6 . . ,
. . .
. ' 3;2
.
R Over.':cai
Ctiiitics 1.30 4.3 ,' . 1;30 ,
4.3 . .
De ?- .. ,. : .c..-..on . ,:
\4 a. ..., -- .-. - - ..-r.ce -
1.19 '
.54
3.9
1.8
1.06
.40' '
3.5, '
-1.6.
Tnsc... . A-.,cc - 4 :>:erest . .A9 2.9 , .80 .: . . 2,7
-
s;;i.- , J4!S - .I5 - .5 - .IS
.5 .
CY,,- LT?cra? ng Costs S11.71 38.R $11.27 ' 37.4
-
3y-7 :occct c red its 2.60) (8.6) (2.60) - (8.6)
yet C:c:ati'r.z Costs
q:c-z; . 'r.vcst-ncnt Sc',irn
$9. ! 1
2.24
30.2 $8.67
- 7.4 2.00
2A. 8
- 6.6
s2*c, ->:.ce, 911.35 37.6 $10.66 35.3
-.
d- :L ?rice,
= 54.00 'tor.:
39.72 31.9 $9.03 29.6
' 7 ~ 1 $j.?O,ton) ,'
Sz.-s: :,
:.
Plant is a 'grass-rcots' facility capable of hahdling run-of*minc cool
from s?ockpile.
LsSor - SS.OO/rnan hour including supervieton and overhead.
2. Utilities - Electricity 3.006/KMrH
Steam Q$ .5O/M Ibe.
4.
Cooling Water B$.O2/M Gallons
Deoreciation - Straight Line over 15 year period.
5. Vaintrnance - 3OL of investment.
6. Insurance & Interest - 5% of Plant Colt.
7. By-product Credits - 8.01 I.pound average for net execs. dirtillate.
8. Pay-Out - R Years. - - - -
/
I

,
!
I- on
'---

THE HEAT OF REACTION?% HYDROGEN AND COAL
A. L. Lee, E. L. Feldkirchner, F. C. Schora,and J. J. Henry*
Institute of Gas Technology
Chicago, Illinois
This investigation is a part of the study conducted at the Institute
of Gas Technology (1,2,4,5,6,7,9,10,11) to obtain the necessary infor-
mation for designing an efficient coal hydrogasification plant. A thor-
mgh literature search reveals that there is no reported data on the
heat 3f reaction of hydrogen and coal. Therefore, two calorimeters were
desisned, constructed, and operated, one to measure the heat of reaction
md the other to meesure heat capacity by the drop method. The heat-of-
reaction calorimeter can be operated at temperatures up to 1500'F and
pressures up to 1500 psia; the drop calorimeter operates at atmospheric
pressure and temperatures up to 1500°F. This paper reports the results
of the following investigations :
APPARATUS
1. The heat of reaction of hydrogen with coals
and coal chars after various degrees of gasi-
fication.
2. The heat of reaction of coa1,pretreatment.
3. The heat capacity of coals and coal chars.
The heat-of-reaction calorimeter built by Dynatech Corporation is
shown in Figure 1. The sample is placed in the upper portion of the
neck which is cold. The calorimeter body is filled with hydrogen and
heated. The temperature of the calorimeter body is measured by foi2
thermocouples and two platinum resistance thermometers. The sample is
lowered into the body after the temperature has remained steady for 2
hours. The change in temperature due to a reaction is then measured as
a function of time.
The drop calorimeter, which was also built by Dynatech, is shown in
Figure 2 . The sample is placed in the top f'urnace until it reaches the
desired temperature. It is then dropped into the copper receiver,and
the heat capacity of the sample is determined from the temperature rise
and heat capacity of the copper receiver.
EXPERIIQCNTAL FESULTS
'J
I' The major problem encountered in determining the heat of coal reactisns
at high temperature and high pressure is prereaction, since coals
1 Ceconpose when heated. Meaningful results can be obtained only if the
A'
czal and hydro&en react at conditions at which the desired temperature
I and pressure are stabilized. Therefore, the method of introducing the
sayple t2 the reaction conditions is critical. The coal must not react
I befsre the conditions are set, and the pressure must not be disturbed
when the coal is introduced.
!:
1 ' * Dynatech Corp. , Cambridge, Massachusetts.
(
>
\
c
f

394
The present method is to keep the sample at room temperature inside
the calwimeter drop tube so that the reaction will not take place prematurely.
Convectim shields are installed to prevent a large heat loss
:rm convectim and to ensure that the.sample is in a cold zone while
L l
L..c cel?rizeter is being stabilized at the reaction cmciitions.
Establishing ki heat balance around the calorimeter was necessary in
order to calculate the heat of reaction from experimental 'data. The
imzt b?Jance is best expressed by:
-
where: AHR = heat of reaction
' !I, =
in
heat input from the neck heater
m = mass
C = heat capacity
P
AT = temperature change
chain
3?c!i term in Equation 1 must be either established by calibration or
acccrately measured. The heat capacity of the calorimeter metal was de-
tr'rixined in the drop calorimeter. These data are shown in Figure 3,
The effective nCp of the reaction calorimeter was calibrated by a ,-*'
constant-heet-input technique in a hydrogen atmosphere at 1000 psia and
4 ,~~.?cratures
q-, from 840° to 1460°F. The results are presented in Figure 4. jy
After the Calorimeter constants were determined, the effective (mCpdT)'s r
of the convection shield, the chain, and the empty sample basket were
czlibrated in the calorimeter to determine the change of heat input wjth
ti-", as shown in Equation 2:
where : K(A) = a constant dependent only on time
4 = time
Combining Equations 1 and 2, we have:
Eq-dation 3 was used to calculate the heat-of-reaction data reported in
this paper.
(1)
\
\s
\

1
I
i Design of a hydrogasification $??ant requires data on the heats of
I reaction of raw coal in the coal pretreatment process, the pretreated
Coal in the low-temperature gasifier, the residue from the low-tempera-
'I twe gasifier in the high-temperature gasifier, and the residue from
/ iA? high-temperature gasifier (4,6). The heat-of-reaction and heat capa-
city measurements are given in Tables 1 and 2. Figure 5 shows the tem-
j perature and pressure profiles of a typical experimental m.
DISCUSSION
>
Until now, the heat of the coal hydrogasification reaction has only
been determined by calculation. These calculations have become more
c Precise as more data became available, but no measurements were made
to check the validity of the calculated data.
J
/ Initially (in the absence of accurate pilot plant yield data), the
, :;eat of reaction was estimated by assuming that coal and carbon were
, ewivalent and that the hydrogasification reaction could be approxi-
:-ited. by (j, 8) :
i' C + 2H2 - CHq
' This approach, of course, is very crude and could not be expected to
give a reliable answer, but it could be usef'ul for comparing the therf
rial efficiencies of various gasification processes.
The next approach, using pilot plant data (7), was to calculate the
'heat of reaction from the heats of combustion of the reactants and pro-
,'ducts. The heats of combustim of various coals could be obtained by
a Parr-bomb calorimeter or calculated by the modified Dulong formula.
'But if one attempts to calculate the heat of reaction of hydrogen and
7 coal from heats of combustion, there axe two drawbacks.
First, be-
)cause the calculation involves taking the differences of large numbers
of the same order of magnitude, chemical analyses of all reactants and
1 products must be very accurate and pilot material balances must be quite
) clme to 100 percent or else the balance must be forced. If this is
i/not the case, large errors can be made in this calculation. This makes
the calculated heat of reaction dependent on the quality of the analy-
' tical data and on the method used to force the balance. The second
'drawback is that the calculated heat of reaction is determined for 25'C,
9 so no information is obtained on the reaction heat at actual reaction
' temperatiires.
v
, of cc',irse, if the heat of reaction could be determined at one temperature,
the heat of reaction at various temperatures could be cal-
Iculated from heat capacity data. Some heat capacity data are available
in the literature for coals and cokes, but none are available for
' the particular coals used here. Moreover, the measurement methods
',u'sed so far are not very accurate. In most cases the gaseous decom-
, position products were allowed to escape from the calorimeters during
I coal hcatup, and, consequently, the heat capacity data are rather
, douhtP,l.
\
2 from the heat of formation data for C + 2H2 - CH4 and the pilot plant
ddata,with that obtained from the calorimetry studies is shown in Fig-
A conprison of the heats of reaction of hydrogen and coal obtained
J WE
i
c. Iiote that the pilot plant data were based on a 77'F reference
,
J

396
‘c erperature, while the present experimental data were obtained at opereti:i;
conditions. Noreover, the experimental data were obtained from
coals at four different stages of reaction: raw coal, pretreated coal,
lw- t emperature gasification residue, and high- t emperature gasification
residue. Ash balances were used to put these results on a common basis.
The ash balance calculations gave the percent of carbon gasified in
each cDal or char. Raw coal was asswed t3 have 0s carbon gasified.
7-L~s Figure 6 shows the general trend of the heat of reaction.
Accurate heat-sf-reaction data are given in Table 1. Although the
pilot plant data are considerably scattered, the average value is not
tm different from that obtained by the other methods. The calorime-
try data also show some scattering, which is due to the hetemgeneous
nature of the mal, and the characteristics of the calorimeter and the
sensiw instruments. Examinations of the temperature measurement, the
pressure measurement, the temperature distribution in the calorimeter,
the total mass balance, and the calibration results obtained from the
cmstant-heat-input method and experimental runs on hydrogen and 2-
aecane reactions indicate that the data reported in Table 2 should not
::zve T: deviation greater than 10%.
This work is jointly sponsored by the American Gas Association and
tile U.S. Department of the Interior, Office of Coal Research. Their
suppopt is gratefilly acknowledged. Valuable advice and discussions
riere provided by Drs. B. S. Lee, S. A. Weil, and C. W. Solbrig of the
Institute of Gas Technology. J. R. WSando assisted in the experimen-
tal :.:3rl.;.

i
Ij
EPERENCES CITED
307
HJ7drOgen Mixtures at Eigh Temperatures and Pressures, " Ind. E~P.
Chex Process Desim Develop. 4, 134-42 (.1965) April.
Huebler, J. and Schora, F. C., "Coal Hydrogasification, Chem.
Erg. Prom. 62 87-91. (1966) February..
,'
' JANAF Thermochemical Tables. Distributed by: Clearinghouse for
Federal Scientific and Technical Information, Springfield, Va.
KaVlACkr V. J. and Lee, B. S., "Coal Pretreatment in Fluidized
Bed, he??. Chem. SOC. Div. Fuel Chem. Preprints 10, No. 4J
131-39 (1966 September.
Lindentl H. R., "Pipeline Gas From Coal: Status and Future Prospects,
Coal Ape 5, 64-71 (1966) January.
Pyrcioch, E. J., Lee, B. S. and Schora, F. C., "Hydrogasification
of Pretreated Coal for Pipeline Gas 'Production, It her. Chem. SOC.
Div. Fuel Chem. Preprints 2, No. 4, 206-23 (1966) September.
Pyrcioch, E. J. and Linden, H. R., ";ipeline Gas by High-Pressure
Fluid-Bed Hydrogasification of Char, Ind. Erg . Chem. E,, 590-94
( 196 0 ) July.
Rossini, F. D. et al., Selected Values of Physical and Thermo-
dynamic Properties of Hydrocarbons and Related Compounds.
Pittsburgh: Carnegie Press, 1953.
Tajbl, D. G., Feldkirchner, H. L. and Lee,IIA. L., "Cleanup
Methanation for Hydrogasification Process her. Chem. SOC.
Div. Fuel Chem. Preprints l.0, No. 4, 235-45 (1966) September.
Tsaros, C. L. and Feldkirchner, H. L., "Production of High Btu
Gas, 'I Chem. -*62, 49-54 (1966) August.
von Fredersdorff, lIC. G. and Vandaveer, F. E., "Substitute Natural
Gas From Coal, in Segeler, C. G., Ed., Gas Wineers Handbook,
Section 3, Chap. 9, 3/100-3/123. New York: The Industrial
Fress, 1965.
I

2,
'!
!
Proximate Analysis, wt $
Moisture
Vo lat i les
' Fixed Carbon ,
Ash
Total
ntimte Analysis, wt %
Cerbon
Hydrogen
Nitrogen
Oxygen
Sulrur
Ash .
Total
Reactyon Temperature, OF
Reaction Pressure, psia
Coal Reacted (Av J, %
Carbon Gasified YAvg), %
Avg AHR, Btu/lb coal
rkacted
3P8
Table 1. ANAE$SIS OF COATl AND COAL CHARS
Raw Pretreated Low-Temp FTigh-Temp
Coal Coal .' Residue Residue
-
1.3
34:6 '
52.0
-&&
0.5
23.3
63.5
d%
0.6
4.6
77.6
1oo.o 17*2
0.6
3.3
71.6
24.5
m
71: 2
5.14
1.23
6.03
70.1 ~
3.70 '
1.37
* '76.9'
2.05
1.01
0.55 ,
4.19
12!21.
100 I 00
1300
1000
$49.7
50.6
181% .
*' 8.30
-. 3.80
-
12 * 73
100.00
1300
1000
47.2
41.0
I
1919 ',
2 .&I
17 30
m37
1300
1000
27.3
29.5
2432
Table 2. HEAT OF REACTfm OF COAL. . \
I
72.6
1.08
0.54
0.00
1.24.
24.62
100..00
1300 \
1000
28.8 ,
26.0,
3078
Temperature, Pres sure, Coal 'Carbob mR,, Btu/lb'
Material OF pia I Reacted,% Gasified,$ coal' reacts
d
Raw coal
1000 41.5 . -- 1300 /
Raw coal
1000 .52.4 50.7 8 1951 \
Raw coal . 1300 1000 52.2 '52.0 1723
RZIT coal ,1300 1000 48.1 55.6 1788 c
Raw coal 1300 1000 46.1 44.2 18 07 1
Raw coal 1400 1000 51.7 52.3 1775
?retreated coal 1300
Pretreated coal 1300
Pretreated coal 1300
Pretreated coal 1300
Pretreated coal 1300
bw-temp residue 1300
bw-temp residue 1300
HiZh-temp residue 1300
r.
1000
1000
1000
1000
4000
1000
1000
1000
7 -
*
46.1 37.6 1935 $\
47.3 41.2 1717
47.3 41.5 1844
47.1 40.0
48.3 45.0 212:; *
\
\
?
\
- \

I
i /
- .
399
. . .
I .
2’
0
H
fi
rl
a)
h
J

0 IS
0 14
LL
9 013
e
\
3
b-
m
*- 012
t
V
a
4
V
b- 0 I1
a
W
I
0 IO
0 09
200 400 600 eo0 1000 1200 1400 1600
TEMPERATURE . OF A-Ile14OS
Figure 3. MEAN HEAT CAPACITY OF CALORIMETER
(From 70°F to Temperatures Indicated)
800 900 1000 1100 1200 1300 1400 600
, TEMPERATUREPF L-IIII404
Figure 4. CALIBRATION OF HEAT-OF-REACTION CALORIMETER

I20 0 zo -0 SO 00 100 110 110 160 In0 200 I10 140 1.0 zoo
TIME, min
Figure 5. TIME-rTEMPERATLTRE DATA FOR HEAT OF REACTION OF
HYDROGEN AND COAL
3900
3700
3wx)
3300
w c
y 3100
Y
22.00
d
0 2700
m
2
2 2500
+
m
,i 2=
0 NOT PLANT DATA
? Zloo b RAW COAL .
RlETREAIED COIL
1000 V LOW-TEMP RESIOVE
0 MIGM-TEMP RESIDUE
1700 MAT-Of-FORMATION
1300
0 IO 20 30 - 40 50 60 m m
CARBON GASIFIED. Y .
,101.s*
Figure 6. COMPARISON OF DATA FOR HEAT OF REACTION
OF HYDROGEN AND COAL
J

402
HYDROGEN CYANIDE PRODUCED FROM COAL AND AMMONIA
, .-.
G. E. Johnson, 'W. A. Decker, A. .J. Forney, and J. H. Field ' ,':'
Pittsburgh Coal Research Center, U. S. Bureau of Mines,
4800 Forbes Avenue, Pittsburgh, Pennsylvania 15213
' INTRODUCTION, .
Hydrogen cyanide (HCN) has been one of the country's strongest growth
petrochemicals in recent years. U. S. production has increased from 174 million
pounds in 1960 to 350 million pounds in 1964, a 100-percent increase over the
4-year period. The growth of production of hydrogen cyanide has been directly
'related to the expansion in production of synthetic textiles from acrylonitrile.
Relative growth and production of hydrogen cyanide and acrylonitrile is as
follows :.
Production of Hydrogen Cyanide and Acrylonitrile
Hydrogen cyanide,?' Acrylonitrile,
Year million lb million lb
1960 174 229- lo/
1961 211 25%;
1962
1963
1964
1965
266
293
350
---
360-
10/
455-
593-
10/
371 (6 months)- 4/
About 50 percent of the total output of hydrogen cyanide goes into the
production of acrylonitrile; most of the remainder is use in production of
adiponitrile and the manufacture of methyl methacrylate.1' However, in recent
years acrylonitrile and adiponitrile are being produced by processes which
generate hydrogen cyanide as a byproduct.?/ The bulk of acrylonitrile is a,
used in production of acrylic fiber (Orlon, Acrilan, Dynel, Zefran, etc.),-
a smaller amount in production of nitrile rubber, the adiponitrile in manu-
facture of Nylon.
The manufacture of sodium cyanide utilizes about 7 percent of hydrogen
cyanide production. The remaining hydrogen cyanide goes to a large number of
relatively small uses including ferrocyanides, acrylates, ethyl lactate, lactic
acid, chelating agents, optical laundry bleaches, and pharmaceuticals.
The Andrdssod process is the major commercial process used for producing
hyjrogvli :/aiide. It invo1v:s the reaction of methane, ammonia, and air over
a platinum catalyst at l,OOOo to 1,200' C.21 The platinum catalyst is usually
alloyed with rhodium (10 to 20 percent).
Conversion by the Andrussow process in a single pass is limited to about
69 percent of the ammonia (about 75 percent with gas recycle) and 53 percent
of the methane. A typical analysis of the reaction gases leaving a catalytic
reactor is as follows in volume-percent: Nitrogen 56.3, water vapor 23.0,
hydrogen 7.5, hydrogen cyanide 6.0, carbon monoxide 4.4, ammonia 2.0, methane
0.5, carbon dioxide 0.2, and oxygen 0.1.
a/ Reference to trade names is made for identification only and does not imply
endorsement by the Bureau of Mines.
I

3
1
403
In the Catalytic Degussa process which is not in general use but is similar
to the Bureau of Mines method in that heat is provided externa.lly, the offgas
from the ammonia-methane reaction contains more than 20 percent hydrogen,cyanide.
Utilization of methane and ammonia are reported as 91 and 85 percent, respect-
ive ly .
In its search for new uses for coal the Bureau of Mines has been .investigating
the production of hydrogen cyanide from cpa.1. Although hydrogen cyanide
is present in coke oven gases, and at one time was recovered as a byproduct,'
this source of the gas has not been commonly used in the United States since the
development of the newer methane-ammonia.processes. Although the production of
hydrogen cyanide from coal is technically feasible, production in yields that
could be competitive is a problem of major concern.
..
. .
EQUIPMENT AND PROCEDURE
.. .
Figure 1 is a flowsheet of the experimental unit. Coal ground to minus 300
mesh is dropped in free-fall through a heated reaction zone in the presence of
ammonia at rates up to 1.10 lb/hr. A revolving-disk feeder especially designed.
by the Bureau of Mines to feed coal at low rates was constructed; it delivered "
to within 5 percent of the desired feed rate.
A special coal feed system is used to prevent agglomeration and possible
plugging of the reactor by heating the coal rapidly through its plastic range
(about 400' C).
The reactor itself is a &-foot length of vitreous refractory mullite,
1-1/8 inches ID and 1-3/8 inches OD, jacketed with two electrical resistance
heaters. The top heater (maximum temperature 850" C) is 12 inches long and is
wound with Nichrome wire. It serves as preheater for the coal and gas. A
Kanthal heater (Al-Cr-Co-Fe alloy, maximum temperature 1,250' C) encloses the
center 20 inches of the tube, or the reaction zone. The bottom section of
the reactor is exposed to the atmosphere for rapid cooling of the product gases.
The bottom of the reactor tube fits into a 4-liter side-arm flask or char
receiver in which the heavier solids are collected. The fine solids and carbon
black produced are collected in an electrostatic precipitator. After the
product gases leave the precipitator they pass through a cooler, then they are
either metered or sent through absorbers to remove the hydrogen cyanide for
analysis.
All of the piping and vessels are stainless steel or glass in order to
counteract the corrosive nature of the gases. Since the gases are toxic, the
unit is completely enclosed, and the enclosure is well ventilated to prevent
any accumulation of escaped gases. The whole structure (6 ft x 6 ft x 15 ft
high) is covered with steel sheeting. It has an exhaust blower (400 cfm) on the
roof and access doors at both ground and 8-foot levels. Figure 2 shows the
exterior of the unit and figure 3 shows its interior.
Product gas can be recycled to the top of the reactor adjacent to the
cooled feed tube along with part of the feed gas. This flushes away any tar
vapors which might adhere to the walls and cause plugging. The remainder of
the feed gas (0 to 10 scfh) enters the reactor with the coal. Ammonia, helium,
methane, nitrogen, or air, or mixtures of these gases fed from cylinders have
been used as feed gas.

404
Before startup, the system is purged with inert gas. After the reactor
has been heated to 1,250' C the desired flows of coal and gas. are started.
Ammonia and nitrogen OK helium are the gases usually used. The gas flow is
generally split, part entering the top of the reactor adjacent to the cooled
feed tube, and the remainder entering with the coal.
The powdered coal is fed through a steam-jacketed tube (5/16Linch OD) which
extends into the preheat zone of the reactor. .The coal leaves the end of the
feed tube which is at the temperature of the steam to enter the preheat zone of
850' C. The temperature of the coal rises very suddenly to 850" C because of
the high heat-transfer rate to the small particles. The carrier gas (usually
helium, an inert gas) fed with the coal keeps the particles in motion and helps
prevent agglomeration as the coal rapidly passes through its plastic range.
Proximate and. ultimate analyses-/are made of the char and heavier
solids collected in the char receiver and of the lighter solids collected by
the electrostatic precipitator. Mass spectrometric and chromatographic analyses
are made on spot samples of the product gas. For cyanide determinations,
metered amounts of product gas are bubbled through two scrubbers in series
containing solutions of sodium hydroxide. Titration with silver nitrate solu-
tion determines the total cyanide pre sent .i/
EXPERIMENTAL RESULTS AND DISCUSSION
The initial tests were made with a metallic reactor tube, but because of
low yields of hydrogen cyanide and failure of the metal at the temperatures
employed, the metal tube was replaced by a ceramic reactor tube.
In all the tests of this report with hvab coal, Pittshrgh seam coal from
Bruceton, Pa., was used. Its ultimate analysis is as follows in'percent:
Carbon 7.5.6, ash 8.4, oxygen 8.0, hydrogen 5.1, nitrogen 1.6, and sulfur 1.3.
The effect of varying the coal-ammonia feed rates is illustrated in table 1.
Hydrogen cyanide yields were computed from the wet-chemical method of analysis
which is considered the more reliable method since it was determined from pro-
portionated gas samples taken continuously throughout the test (sample volume
of 0.2 to 2 cu ft). Only spot gas samples (sample volume 0.01 cu ft) were used
for the chromatograph and mass spectrograph analyses.
In test C-241 the'coal-feed rate was 0.37 lb/hr and the ammonia-feed rate
was 1 cu ft/hr. (All process gas volumes reported are corrected to standard
.conditions of 0" C and 760 mm mercury pressure.) A hydrogen cyanide yield of
0.6 cu ft per cu ft of ammonia reacted was obtained, corresponding to about
12 percent hydrogen cyanide in the product gas.
In test C-243 the hvab coal-feed rate was,reduced to 0.19 lb/hr, while the
ammonia-feed rate was increased to 2.3 cu ft/hr. A yield of 0.4 cu ft hydro-
gen cyanide per cubic foot of ammonia consumed was obtained, corresponding to
about 13 percent hydrogen cyanide in the product gas.
The yield of hydrogen cyanide per pound of coal was approximately doubled
when the coal-feed rate was halved and the ammonia-feed rate doubled (tests
C-241 to 243), while the hydrogen cyanide yield per cubic foot of ammonia con-
sumed decreased by one -third.

’f ’f
\
?
i
i
k
>
/
f
j
I
405
Table 1.- Data from Tests with hvab Coal and Ammonia with Helium at 1,250° C
Product gas, percent
Test HCNL’ H,S N& H2 O-, N, CO ’ C02 C& CZ% C3H8 C8H10
C-241 11.8 0.0 0.0 77.7 0.1 6.7 9.6 0.3 5.1 0.4 I 0.1 0.0
C-243 13.1 .O 17.5 68.6 .9 6.8 4.0 .1 1.0 .o .o 1.1
C-239 13.2 .1 .5 75.9 .1 4.4 10.6 .5 6.9 .6 .I .3
C-198 5.4 .3 .O 78.3 .3 3.5 11.1 .3 5.7 .5 .o .o
Feed Yield, cu ft
Total
Coal off gas Length HCNIC~
HCNICU
He 3 NY3r feed, He-free, of run, HCN/lb ft N& ft N b
cu ft/hr cu ft/hr lb/hr cu ft/hr min coal feed consumed
C-241 1-00 1.00 0.37 5.06 30 1.59 0.60 0.60
C-243 0.49 2.35 .19 4.90 30 3.31 .27 .40
C-239 .98 1.00 .73 5.59 15 1.01 .74 .76
C-198 2.12 0.20 .37-.40 3.61 15 0.51 .98 .98
- 11 Wet-analytical method of analysis for HCN only; other components are the
average of 2 chromatograph and 2 mass spectrometer analyses, HCN-free basis.
In test C-239 (table 1) the coal-feed rate was increased to 0.73 lb/hr,
while the ammonia flow was maintained at 1 cu ft/hr. A hydrogen cyanide yield
of about 0.8 cu ft per cubic foot ammonia reacted was obtained, equivalent to
about 13 percent hydrogen cyanide in the product gas.
In the next listed test, C-198, the coal-feed rate was 0.37 to 0.40 lb/hr,
while the ammonia flow was only 0.2 cu ft/hr. The hydrogen cyanide content in
the product gas was only 5 percent, table 1, but conversions of about 100
percent of the ammonia were obtained with 1 cu ft of hydrogen cyanide formed
per cubic foot of ammonia used. This yield of hydrogen cyanide approximates
the stoichiometric yield according to the following reactions:
C + N& - HCN + H2
C& + HCN + 3H2.
In typi cal commercial units using catalysts for the methane-ammonia reaction,
the ammonia conversion attained is about 75 percent. Yield values
include .the slight amount of hydrogen cyanide that may be formed when coal is
heated to hi gh temperatures without adding ammonia.
In general, the tests of table 1 indicate that an excess coal feed is
desirable in order to attain maximum utilization of the ammonia since the
ammonia is by far the more expensive raw material.

4 OL.
'
,
I.
.
TesCs. With.Coals .of Different Rank
. . .
In addition to the h,yab -coal, lignite, subbituminous, low-volatile bituminous,
and anthracitc coals. coal char, and activated carbon were' tested as
raw materials for producing hydrogen cyanide. It is thought that the volatile
matter in.coa1 reacts with the ammonia to form hydrogen cyanide, therefore
coals with higher volatile content should produce more hydrogen cyanide.
The
volatile-matter contents on a moisture-free basis of the various materials
tested are as follows:
Volatile
matter,
Tes t Identification of coal Source of coal percent
C-123, 124,. Hvab, Pittsburgh seam Bruceton, Pa. 34.0
125, 198,
239, 241,
243
C -207 Activated carbon, Union Carbide an,d 2.0
Grade SXWC Carbon Corp.
c -2 10 Anthracite Anthracite Research 7.6
Center, Schuylkill
Haven, Pa.
C -246 Subb, Laramie seam Erie, Colo. 38.5
C -249 Lvb, Pocahontas #3 seam Stepheson, W. Va. 17.5
C-252 Lignite, unnamed seam Beulah, N. Dak. 41.1
C-254
1/
Pretreated hvab,-
Pittsburgh seam
Bruceton, Pa. 32.6
- 1/
Treated with air at ZOOo C.
Table 2 shows the results of these tests. Lignite with 41 percent volatile
matter produced the most hydrogen cyanide, 0.4 cu ft per cubic foot of ammonia
consumed; activated carbon, containing the least volatile matter (2.0 percent),
produced the least hydrogen cyanide, 0.007 cu ft per cubic foot of ammonia
consumed.
The chemical nature of the volatile matter and the oxygen content of the
coal may also affect the hydrogen cyanide yield. The ratio of H, to CO in the
off gases varied from 2.3 to 1 for subb coal and 2.4 to 1 for lignite to 7 to 10
to 1 for hvab. The higher carbon monoxide values obtained uith subb coal and
lignite are due to the higher oxygen contents of these coals, being 17.1 and 20.3
percent, respectively, compared with 8.0 percent for the hvab coal.

I
b
1
I
Table 2.- Data from Tests with Various Coals and Ammonia with Helium at 1.250" C
I
I Product gas, percent
I Test HCNL' H2S & H2 02 N2 CO C02 Ct4 C2H4 C3H8
I
' C-246 Subb 5.4 0.0 0.0 64.2 0.0 4.9 28.0 0.5 2.4 0.0 0.0
, C-249 Lvb 4.2 .O .O 78.9 .1 14.2 3.6 .O 3.0 .2 tr.
C-252 Lignite 6.4 tr. .O 64.6 .1 6.1 26.3 .5 2.4 .O .O
C-254 Pretreated, hvab 3.8 .1 .O 72.3 .8 10.6 14.3 .O 1.9 .O .O
j C-210 Anthracite 0.2 .o .o 73.4 .3 21.4 2.9 .4 1.5 .o .o
C-207 Activated carbon .25 .O .O 70.0 .O 23.0 6.6 .1 0.3 .O .O
9 Feed
i
I
He 9
cu ft/hr
NH3 9
cu ft/hr
Sol ids
feed,
lb/hr
Total
off gas
He-free,
cu ft/hr
'Length
of run,
min
1 C-246 Subb 0.91 1.00 0.42 6.88 30
C-249 Lvb .97 1.00 .47 3.92 30
C-252 Lignite 1.06 1.00 .37-.40 6.38 15
C-254 Pretreated, hvab 1.04 1.00 .35 5.80 30
C -210 Anthracite 1.02 1.15 -37-.40 4.15 15
f C-207 Activated carbon 1.06 1.15 .62 3.23 15
/ Yield, cu ft
1 HCN/cu ft HCN/cu ft
J HCN/lb coal feed consumed
,I C-246 Subb 0.88 0.37 0.37
C-249 Lvb
/ C-252 Lignite
! C-254 Pretreated, hvab
C-210 Anthracite > C-207 Activated carbon
.34
1.05
.63
.02l2/
.013-
.16
.41
.22
.007
.007
.16
.41
.22
.007
.007
I
)
L/
- 21
Wet-analytical method of analysis for HCN only; other components are the
average of 2 chromatograph and 2 mass spectrometer analyses, HCN-free basis.
Per pound of carbon.
Tests were made to determine the effect that oxygen in the treating gas
would have on the yield of hydrogen cyanide. It was thought that the heat for
raising the temperature of the reactants to reaction temperature could be supplied
by direct contact with a hot flue gas containing oxygen instead of by electric
heating (table 3). In test C-125, a maximum yield of hydrogen cyanide was pro-
duced for tests with air in the treating gas--0.12 cu ft of hydrogen cyanide per
cubic foot of ammonia consumed, or 9.2 percent hydrogen cyanide in the product
gas. When more than 0.5 percent. air was fed into the reactor, moisture condensed
on the walls of the solids-collection flask; the yield of hydrogen cyanide de-
creased. The moisture could have been absorbing the hydrogen cyanide since
hydrogen cyanide is highly soluble in water. This approach was abandoned.

Table 3.- Product Gas Analyses and Yields of Hydrogen Cyanide from
Tests with hvab Coal. Ammonia, and Air at 1,250" C
Product gas, percent Length Feed
of run, N&, Air, Coal,
Test HC&' C& NH3 H2 N2 CO min cu ft/hr du ft/hr lb/hr
C-123 10.3 1.1 10.2 62.8 16.2 9.7 20 5.90 0.52 0.17-0.20
C-124 10.9 0.6 6.9 62.1 17.2 13.2 20 2.96 .53 .17- .20
C-125 9.2 .4 9.6 48.4 31.5 10.1 20 3.08 1.08 .17- .20
Total Yield, cu ft
. off gas, HCN/cu ft HCN/cu ft N&
cu ft/hr HCN/lb coal NKq feed c on s ume d
C-123 4.18 2.30 0.073 0.078
C-124 3.05 1.77 .112 .120
C-125 3.70 1.82 .110 .123
- 1/ Wet-analytical method of analysis for HCN only; other components are the
average of 2 chromatograph and 2 mass spectrometer analyses, HCN-free basis.
Tests With Methane and Ammonia
Tests were made without coal feed %ut with amnonia and methane to determine
the resulting hydrogen cyanide yields for comparison. In one series of tests
2.0 cu ft/hr of methane was reacted with ammonia in flows varying from 0.3 to
2.5 cu ft/hr at 1,250' C. As illustrated in figure 4, yields of 0.18 to 0.61
cu ft of hydrogen cyanide per cubic foot of ammonia consumed were obtained,
equivalent to 1.2 to 13.8 percent hydrogen cyanide in the product gas. The
yield of hydrogen cyanide reached a maximum at a feed ratio of methane to amonia
of about 1 to 1.
' A series of tests was made in which the methane and ammonia flow rates
were maintained at 2 and 1 cu ft/hr, respectively, while the reactor temperature
was increased from l,OOOo to 1,275' C. Hydrogen cyanide yields are plotted
with temperature in figure 5, indicating increased hydrogen cyanide yields with
increased temperature. Maximum yields of about 0.6 cu ft of hydrogen cyanide
per cubic foot of ammonia consumed were obtained at the higher temperatures,
while at 1,000" C only 0.07 cu ft of hydrogen cyanide per cubic foot of amnonia
consumed was formed.
To explore the use of coal-derived gases in the formation of hydrogen
cyanide from amnonia, synthetic mixtures of a low-temperature carbonization
gas, a coke oven gas, and a producer gas were prepared and reacted with amnonia
at 1.250" C. Gas flows were adjusted to give a minimum methane-to-ammonia
ratio of 1 to 1. In figure 6 the hydrogen cyanide yields obtained are plotted
with the methane' contents of the coal gases. Increasing hydrogen cyanide yields
were produced with increasing methane contents of the feed gas.
\
\

409
ECONOMIC EVALUATION
The Bureau of Mines Process Evaluation Group: Morgantown, W. Va., made a
preliminary cost study of an integrated plant to produce hydrogen cyanide
by reaction of ammonia with coal. The cost study was based on experimental
results including a yield of 0.6 cu ft of hydrogen cyanide per cubic foot of
ammonia. Electrical heating was assumed as in the becch-scale tests; a plant
capacity of 40 million pounds per year was chosen. The total estimated
capital investment was $12,930,000 including costs for power generation.
Based on a coal cost of $4.00 per ton and an ammonia cost of $100.00 per
ton, the operating costs before profit and taxes would be 5.82 cents per
pound of hydrogen cyanide product allowing byproduct credit. Addition of
12-percent gross return on investment would give production costs of 9.7 cents
per pound of product when $4.00 per ton coal is used. The current market
price is 11.5 cents per pound.81
Credit has been allowed in the cost figures for a 7.6-percent yield of
carbon black and the excess char produced in the process. Some of the char
and the scrubbed product gas (containing about 75 percent hydrogen) are
consumed in the steam plant for power generation. Electrical heating, which
was used in the test unit and also in the cost figures, is one of the most
expensive types of heating, accounting for greater than 40 percent of the
capital costs in the estimate. If cheaper conventional heating could be used,
production costs would be lowered considerably.
CONCLUSIONS
Hydrogen cyanide has been produced from coal and amnonia at 1,250" C in
bench-scale studies. The use of a metal reactor was unsuccessful because the
metal failed at the temperatures required, and the yield of hydrogen cyanide
was low. The yield was improved greatly when a refractory ceramic reactor was
used.
Hydrogen cyanide yields approximating stoichiometric of 1 cu ft of hydrogen
cyanide per cubic foot of ammonia reacted were obtained at low flows of ammonia.
At higher amnonia flows, ammonia conversion of about 75 percent was obtained,
which is the usual conversion attained in commercial units using natural gas
and a platinum catalyst.
The low-volatile coals gave low yields of hydrogen cyanide; the high-
volatile coals gave the best yields. The results indicated that the hydrogen
cyahi.de is produced by reaction of amnonia with the hydrocarbons in the coal.
Yields of hydrogen cyanide from reaction of ammonia with gas mixtures con-
taining methane are directly related to the methane content of the gas.
Cost studies indicate that hydrogen cyanide can be produced from coal
and aunnonia at a price approximating the posted sales price.

410
BIBLIOGRAPHY
1. American Public Health Association, Inc. Standard Methods, for the
Examination of Water and Wastewater. 11th ed., 1960, pp. 355-,356.
2. A.S.T.M. Standards with Related Material. Gaseous Fuels, Coal and Coke.
Part 19, 1966, 470 pp.
3. Chemical Week, Market Newsletter, v. 99, No. 23, Dec. 3, 1966, p. 65.
4. Chemicals, Quarterly Industry Report, U.S.. Business and Defense Services
Admin., v. 12, No. 4, Dec. 1965, p. 40.
5. Current Industrial Reports, Inorganic Chemicals and Gd'ses 1964. U. S.
Commerce Department, Series M-28A, Released Jan. 28, 1966, p. 6.
6. Fieldner, A.C., and W. A. Selvig.. Met!iods of Analyzing Coal and Coke.
BuMines Bull. 492, 1951, 51 pp.
7. Fugate, W.O. Hydrogen Cyanide, pp. 574-585 in v. 6 Encyclopedia of
Chemical Technology. Interscience Publishers, New York, N.Y., 1962,
2d ed., ed. by R. E. Kirk. and D.O. Othmer.
8. Oil, Paint and Drug Reporter. V. 189, No. 23, June 6, 1966, p. 20.
9. Sherwood, Peter W. Synthesis of Hydrogen Cyanide by Autothermic Reaction.
Refining Engineer, v. 31, No. 2, February 1959, pp. C-22-26.
10. U. S. Tariff Commission, Synthetic Organic Chemicals, U.S. Production
and Sales:
1960 - Tariff Comission Publication No. 34, p. 53, published 1961.
1961 - ditto 72, p. 53, published 1962.
1962 - do. 114, p. 56, published 1963.
1963 - do. 143, p. 55, published 1964.
1964 - do. 167, p. 55, published, 1965.
All'published in Washington, D. C.
\
,
1
i

W
0

Figure 2. Enclosure surrounding hydrogen cyanide unit.
1
‘1

Figure 3. Hydrogen cyanide reactor.

0 0
f v)
C
0
u
m
I
z
2
c
2
r
d
J
w
*
z
V
I
0.f
5 ..
.4
7 .-
.2
.I
CG-6
- _ ..
I I
44
C%-3
/
0 0.5 I .o 1.5 2 .o
NH, FLOW,scfh
12 -7- 66 L-9726
I

C
:G-l I
I
00 1,100 1,200
TEMPERATURE, OC
1
0
CG-3 /
12-8- 6 6 L-9727

0. E
c
.4
7 ..
.c
. I
(
-
Low temperature
carbonization gas
/
I I
20 40 60 80 IO0
CH, IN FEED GAS, percent.
I2 -7- 66 L-3728
\
a
\
i

417
h6SSBAUER SPECTROSCOPY OF IRON IN COAL
John F. Lefelhocz*, Robert A. Friedel' , and Truman P. Kohmd
*Depafiment of Chemistry, Carnegie Institute of Technolow
'Pittsburgh Coal Research Center, Bureau of Mines, U.S. Department of the Interior
Pittsburgh, Pennsylvania 15213
INTRODUCTION
Metallic elements (Na, Mg, Al, Si, K, Ca, Ti, Fe) which occur abundantly in
the ash obtained fron the combustion of coals are present in the original coals
partly as inorganic constituents or minerals. The identification of minerals
present usually does not account quantitatively for several of these elements, and
it is supposed that some sort of organic bonding with the coal is involved. The
nature of this bonding in most coals has not been determined. Changes in the infrared
spectra of lignites and brown coals following acid and alkali treatment indicate
the presence of metallic salts of carboxylic acids (1,2,3). For higher rank coals,
iittle or nothing is known about the siructures of the metallic elements
organicalljr bound". In the case of iron, for example, it has not been determined
whether the so-called "organict1 iron is in the ferrous or ferric state, to what
Element fr elements the iron is bonded, or whether there is more than one form of
,
organic iron.
The term "organic" must be interpreted with care; the nature of
the bonding is uncertain, and presumbly could be ionic, coordinate, or organo-
metallic.
Mineral components oontaining iron can often be identified petrographically.
Pyrite (FeS2) is common, both as distinct nodules and as veins, often intertwined
with carbonaceous macerals. An electron microprobe study of several coals (4)
revealed, however, many examples which lacked a correlation between Fe and S
distributions. In some cases Fe was distinctly associated with Si and AI.,
suggesting incorporation in an aluminosilicate gel or kaolinite. A strong
parallelism of Fe and Ca (but not Si, Al, K, or S) in one specimen suggested the
presence of Fe as carbonate or possibly oxide. In some cases the Fe shared
fairly uniform distribution with no ftpparent correlation with other ash-forming
elements. This presumably could be organic" iron.
M'dssbauer spectroscopy has been employed to study the chemical properties of
iron in a great variety of natural materials: oxides and oxyhydroxides (5,6),
sulfides (7,8) , numerous silicate minerals (5,6,9,10,11,~)~. ilmenite and related
titanium minerals (5,13), siderite (6), jarosites (5,14), lollingite (15); ordinary
chondrite meteorites (16) , carbonaceous chondrites (13, an achondrite (16), and
several tektites (11).
Recent general articles on Mzssbauer spectroscopy (18,19,20,21,22) discuss
the interpretation of spectra in terms of the chemical state of iron, including
its oxidation state, bonding, and environmental symmetry. The isomer shift (d),
quadrupole splitting (A), line width (r), line intensities, and a comparison of
the spectra obtained at room temperature and at liquid-Ng temperature are useful.
We have undertaken Mgssbauer studies to characterize non-mineral iron in

418
cual. A mjor aa\;antage or' the bf6ssbauer technique is that the iron is observed
without alteri:ig its chemical state or eniiiroluuent. The relatively low atonic
nilnber Oi' the mjor deinents present :Bakes it possible for small: amounts of iron
to >e obser.;ed in rather large amoun':,s 01' cml. Sa:nples of whole coal and vitmin
as free as 2ossible from ninerals xere selected in wder to iyiinimize mineral-iron
interzerence, and a I'ew saiqles vhose chemical analysis indicated "organic" iron
were included.
I
ES'ERIMENTAL PROCEDURES
Nine specixens, yritrain or vhole-coal, of seven coals with rank from lignite
(72:. C) to anthracite (93;: C) were ased. Their designations and geographic origins
are given in Table 1, in order of increasing carbon content.
Table 2 is a compilation of the analytical data for C, H, N, S, Fe, ash, and
0 (by difference) on samples of these materials, as determined by the Coal Analysis
group at the Bureau of Mines. The total iron in the coal was determined by ash
analysis. HC1-soluble iron was determined by treating a separate powdered sample
with 23;; HCl to extract any iron present as carbonates, oxides, sulfates, etc.
Pyrite iron was then removed by treating the HC1-leached coal with 25;; €IN& to
dissolve the iron conbined with sulfur; the extract was evaporated to dryness to
expel oxides of nitrogen, and the residue was dissolved in HC1. In each case the
iron was reduced with SnC12, the slight excess of which was eliminated with HgC12.
"he reduced iron was titrated with potassium dichromate. "Organic" iron was then
calculated by subtracting the two acid-soluble iron contents from the total iron.
The Zesulting iron contents are listed in Table 2 as: total, HC1-soluble, pyrite,
and organic .
MEssbauer spectra were obtained on coal samples that were ground in a mortar
and pestle. A cylindrical cell, open on one end with a circular aperture of
7.14 em2, was made by gluing a sheet of lucite 1/52" thick with Duco cement to
the bottom of a circular wall mde from a paper card. A 3.50-g sample was then
placed into this cell. The top window, also 1/32" lucite, was then cemented in
place. The sample mass per unit area was thus held oonstant at 0.4% g cmm2.
Inorganic iron compounds and mineral samples were ground and mounted either
between two lucite sheets held together with Duco cement, or by mixing the solid
with acetone and Duco cement and alluwing this mixture to harden on a lucite sheet.
This latter method was used only when the sample would not interact with acetone
or the cement. The area for these samples was also 7.14 cm2. All of the materials
used in this investigation contained natural iron with presumably 2.19 atom $ FeS7.
The M'dssbauer spectrometer incorporates a Nuclear Science and Engineering
Corporation Model-B lathe-type drive modified in this laboratory for automatic
operation. The operating mode employs constant velocit advanced in increments
of C.05 mm s-'. The detector of the 14.4-kev Co57 -. Fegf gamm radiation is a
Reuter-Stokes proportional tube containing lo$ methane in krypton, feeding through
a single-channel analyzer set at 11 - 17 keV. Absorbers were mounted on the
moveable table pe endicuLar to the radiation beam.
The stationary source consists
of - 5 di of co5Fdirfused into chromium metal. A Bird-Atcrmic scanning count
integrator and Varian chart recorder plot the number of counts in a constant time
interval at each velocity in succession.
Powdered samples of sodium nitroprusside gave an isomer shift, relative to
the CoS7-Cr source, of -0.11 mm s-l, and a uadrupole splitting of 1.68 m &I-'.
With this source a line width of 0.25 mm s-' was obseNed with an absorber containing
5 ny cmS of Fe as K4Fe(CN)e.3&0. The estimated uncertainty in the velocity
setting is about 2: of the velocity and the estimated uncertainty in derived
spectral parameters is - 0.05 mm sL.
The counting time at each velocity was - 5

I n
\D
r-
m
d
c
d
rl
rl
n .
aJ
2
C
W
c,
d
X . a
d
t-
0
2
Po
.>
E.
r: .
2
c,
d
>
r:
aJ
c,
E
d
PI
? ?
0 0
no\
(d'
?
5
P
d
F O x n
(v
k
4
2
d
u
0
c,
c
I1
?I
t

420 Ir
?Y ? ? 9 9 9
pc rod n o c o o 0
Y
0 00 M O
. o
0
-f.
H ' O 9
0 I O I O
0 0 0
0;'
N
c9
0
9
rl
;?
N
9
rl
m
0.
rl
f co'O
? E??
0 0 0
f n o
9 ?Y
0 0 0
O F lnd O d
'?? ?cu. ? ?
0 0 0 0 0 0
I

L.
421
minutes, yielding c 600,000 counts and a relative standard deviation < 0.135.
The velocity region scanned for each sauple was from -2.OC mm s-l to about
+ 4.00 mm s-'.
Most measurements were made a rooi; temperature. A cryostat mde from
I' S t y r O f a $as mounted on the moveable table 0: the instrument ;or low-terqerature
I measurements. The sample was placed in this and submerged in liquid N2. The
, spectrum of each coal sample was run twice at room temperature, befcke and after
, the spectm was taken at liquid42 temperature.
Spectral parameters were generally read from plotted spectra. The data from
sample F(a) were fitted to a double Lorentz curve by a least-squares program using
J an IB1 7C40 computer. Isomer shifts are reported (in m s-') with respect to the
, ismer shift of sodium nitroprusside. Quadrupole splittings are reported (in mm s-')
as the velocity differences between the minima of two associated absorption lines.
I DATA
/
Room-temperature spectra for the nine coal samples are illustrated in Figure 1.
J Each spectrum shows neither, either, or both of just two components: (1) a close
doublet similar to those of pyrite and mrcasite, and (2) a wide doublet similar
to those of many ferrous compounds. Table 3 lists the Mossbauer paramters,
including the fractional peak absorptions, for what we will call respectively the
pyrite and non-pyrite resonances. Isomer shifts and quadrupole splittings obtained
/ with our instrument on sane powdered iron campounds and minerals which were
regarded as possibilities for the non-pyrite spectrum appear in Table 4.
J
/ The parameters observed for the pyrite iron are: isomer shift (d) = +0.54
mm s-l; quadrupole splitting (A) = 0.58 m s-I. All of the coals having a non-
' pyrite absorption as shown in Figure 1 have: d = +l.38 mm s-'; A = 2.62 mm s-'.
p' The computer analysis of sample F(a) indicated both lines in the spectrum had
c widths (r) of 0.39 mm s-l; their intensity ratio is within 57; of unity.
I
i Liquid-N2 spectra showed the same absorption peaks as at room temperature,
f with d = +0.65 m s-l and A = 0.58 mm s-l for pyrite and d = +1.47 mm s-' and
> A = 2.78 mm s-l for the non-pyrite iron.
INTERPRETATION
i
Y The isomer shift is a function of nuclear properties and the electron density
at the absorbing nucleus relative to that of the source or a standard absorber,
such as sodium nitroprusside (23,24). As the electron density at the iron
nucleus, due essentially only to s electrons, increases, the isomer shift decreases
'[algebraically. Thus, ferrous compounds have a more positive isomer shift than
ferric compounds, as the additional 3d electron of the former increases the
d shielding effect on the 3s electrons and thereby decreases their density at the
f nucleus (18). For ferric iron, isomer shifts (relative to sodium nitroprusside)
\ have been observed in the range 4.1 to +1.1 mm S'l, and for ferrous iron from
-0.1 to +1.6 mm S-'.
i;
I
Quadrupole splitting into a two-line spectrum occurs when the iron nucleus
he field consists of two parts, (1) that
f is in an asymmetric electric field.
i produced by the electrons of the iron atom, including those shared with adjacent
1 atoms, and (2) that resulting from charges of the surrounding atoms; each part
J can contribute to asymmetry at the nucleus. Ferric compounds shm quadrupole
splittings from zero to about 2.3 nun s-l; here the asymmetry is caused chiefly by
charges of the surrounding atoms. Ferrous compounds show larger values, frm zero
Ferrous iron can exist in either a high-spin or a low-spin
J
1
i
I
r
I
to - 3.3 mm s-I.

422
Table 3. Room-texperature M'dssbaaer spectral parameters for coal samples"
-
?,TiL c iron I'Jon-Pyrite iron
Saqle 1so:ner Quadru2ole Fractional Is omer Quadrupole Fractional
No. shift I s2licting absorption shift t
abs orpt ion
w&iyy
(Ilm s-1 j ('L"1 s-1 ) 0 (m 3-1) (:?I
A (a) - - N.O.* - - N.O.*
B (a) C.->, ./ 1.5 - - N.O.
c (3) - -
N.O. 1 *39 2.65 2 -3
(b) - N.O. 1.3LJ 2.63 2.1
- - N.O.
D (a) G.;l 0.>5 /.J
(b) 0.23 0.4. 3.6 - - N.O.
E (a) 0.3G 0.32 22.7 1.41 2.70 2.6
(b) 0.A G.3 2 -3
- - N.O.
(c)
F (a)
(b)
o.>c
-
-
0.,3
-
-
1.7
N.O.
N.O.
-
1.38
1-39
-
2.62
2.65
N.O.
11.0
5.5
\
\
\
G (a) - -
N.O. 1.38 2.65 1.5
(b) - - N.O. 1.38 2.65 1.0
H (a) 0.54 0.60 1.3 1-39 2.65 1.5
(b) 0.59 0.66 1.7 1.36 2.60 1.7
- - -
I (a) - N.O. N.O.
(b) 0.56 0.65 1.0 - N.O.
* Weighing 3.50 and distributed evenly over -(.14 em2.
May include marcasite.
t With respect to center of sodium nitroprusside spectrum.
i- Zero-point uncertain by 0.10 m s-'.
t N.O. = not observed; in general, < O.y{.
..
i
\
\
\
\
!
\
\

I 1 1 \
......... ...U ...... .."."..-*....'-... .U!.:; ........ .............. a;. ...... (a) *
. .
. .............. ................
...........................................
;(a)
.. ....
.... ................................... .............
'(a) ............. ... .
..
....
.....
........ . .'
)(b) .......................... ............. ........ '.'.: .............. I.;.
.......
..................
.......
.. ... ..
...
' . .."*. '.
*.
:(a)
.....
.'
' *
..
..
,- ........... -4. .........
c(b)**...,.**, .*.... ...... ............... . -..- ..... ..:.. .........
... ......
3( c) . ..-..-- ... r. ................ .............. ....,\. " ...................... . ............ ':."'...... ...
qa) . :. ... .**'..'...... ................ .e.'.* .
.. e.
o(a) .- .a, . .............
. .
.* .* .... .'
. '
9 .
. .
* .
9 .
............................... .... *...........
.......... . ";* ...,... ....
....... .*L ......" ..... .. ...... ...... ................. . ..................
.'.a ., ,.e
ab) '*"'"' ' -
9. .*. .,.+ 99.- .
.....
I (*) ................. ............., ........ ......... .:*.e :-* . .....
-2
1 I
I I
vgIx)cITY (mm e-')
. ...*.-....*
.. . ...-.* :*.:9.*,
Figure 1. M'ksbauer spectra of coal samples. In maet -see, sach point
represents - 600,000 counts. The m c e was Cos' diffused into chromlm
metal, To convert the velocity scale to the sodium nitroprusside scale,
add 0.11 mm 8' to the indicated values.
I
I
b
1

424
state, depending on the strength of the ligand field around the central iron
atom. In the high-spin configuration the large asymmetry of the valence shell
results in considerable asymmetry at the nucleus; low-spin ferrous compounds have
smaller field asynrmetries, which are due chiefly to the external environments only.
Pyrite and mrcasite, respectively stable and metastable Tom of FeS2, are
low-spin ferrous compounds. Isomer shifts and quadrupole splittings observed in
this laboratory (Table 4) are in agreement with those of Temperley add Lefevre (8).
The parameters of the two minerals are so similar that the spectra cannot be
resolved when both are present. A comparison of the data in Tables 3 and 4
confirms the conclusions fran petrographic (25) and x-ray diffraction (26) studies
that the iron sulfide in coal consists minly of pyrite.
From the Mksbauer data on the non-pyrite iron Observed in several coal
samples (Figure 1 and Table J), it can safely be concluded that this iron is in a
high-spin ferrous state. Furthermore, the combination of such high values of
both isomer shift and quadrupole splitting has been reported only in octahedrallycoordinated
high-spin ferrous compounds, so it is highly probable that the nonpyrite
iron in coals is in octahedral coordination. Neither the analytical nor
the Mksbauer data yields a distinction between inorganic or orgsnic minerals or
compounds.
The natural line width (r,) of the FeS7 gamma radiation corresponds to a
velocity dlfference of 0.0973 mm s-' (27), and the minimum observable line \width
(I') in a MGssbauer spectrum is twlce this, 0.195 mm s-'. Inhomogeneities in the
source and absorber, instrumental "noise", unresolved quadrupole splitting, and
atomic spin-spin relaxation effects (22,28,29) increase the apparent line widths,
and further widening occurs with thick absorbers (30). The minimum width ye have
observed with our instrument is 0.25 mm s-l for a very thin absorber.
The value
of r = 0.39 mm sa observed for the strongest non-pyrite iron spectrum (sample F(a))
indicates that this doublet is caused by a fairly well-defined iron compound or
mineral, though some inhomogeneity may be present.
A large number of iron compounds have strong M&sbauer absorptions (large
recoil-free fractions) at room temperature. Some ferrous compounds show little
or no absorption at roan temperature, but at liquid42 temperature the intensity
is strongly enhanced (31J2). We have observed tNs effect with ferrous stearate,
which must be cooled to liquid-N, temperature before resonant absorption is
detected. Eerzenberg and Toms (5) have observed that samples of y-Fe;?% and
d-FeOOH give nonresonant absorption at roan temperature. These would probably
exhibit an effect at liquld-N2 temperature. Lack of such an effect in the coals
is interpreted to man that there are no compounds present in significant amounts
that do not have appreciable resonant absorption at room temperature.
Isomer shifts and quadrupole splittings depend on the temperature. A
second-order Doppler effect (22) decreases the isomer shift 8s the temperature is
increased. Quadrupole splittings for high-spin ferrous canpounds are affected
by temperature much more strongly than those of other iron ccmpounb because the
population of the dc levels of iron in an octahedral field is determined by a
BoltzlarM distribution (18,22). For example, FeS04"&0 shows a change of d
from +1.53 to t1.56 mm s-l and of A frm 3.19 to 3.47 mm s" in going from roam
to liquid-nitrogen temperature (18).
The observed change in the non-pyrite iron
spectrum in car1 of d from +l.38 to +1.47 mm 6-l ana of A fran 2.62 to 2.78 m sa
is in agreement with our assignnent to this class of compounds.
In some iron compounds where the M&mbauer absorption ordinarily shows a
6-line hyperfine structure as a result of a magnetic field at the nucleus, very

;
i 1
1
1
finely conminuted samples show instead a two-line pattern at room temperature as 8
resilt of them1 disruption of the cacroscopic magnetic domains. KKndig et al.
(33) observed this effect in a-Fe203 in particles of - 50-8 diameter, but at
1iqdid-N~ teqerature the b-line pattern was observed. The absence of any magnetic
splitting in the coal spectra at roo-or 1iq~id-N~ temperatures probably rules out
the possibility that the non-pyrite doublet is caused by a magnetically ordered
msterial (such as Fe203, Fea04, FeC3, FeS, or metal) present as very1 small particles.
, Inequality in the intensity of the components of a doublet my result frm
the ardsatropy of the absorption cross section relative to the crystal axes when
(1) a single-crystal absorber is oriented prererentially with respect to the
d optical axis, or (2) there is snisotropy in the recoilless fraction of the split
312 state; the latter condition results in unequal absorption even with powdered
' samples (Coldanskii effect) (20,21). Most of the samples used in this work were
'
powdered, so only the second effect could be operative, except for biotite, which
was mounted as a sheet. The essential equality of intensity of the two nonpyrite
coal spectrum lines would eliminate any compounds showing unequal absorption
I in powdered samples as possible causes of this absorption.
( At the present state of Mgssbauer spectroscopy, the identification of an
unknown compound from its spectrum can only be made by finding a ham compound
havlng the same spectrum and temperature dependence. We have been unable to find
1 any published combination of spectral parameters, for either natural or synthetic
iron compounds, which match that of the non-pyrite iron in coal. Ankerite
,I
(Table 4), which has not been reported previously, likewise does not match.
I '
In Figures 2 and 3 we have plotted A versus d for all single- or two-line iron
spectra for which we have been able to obtain data (5,6,~,9,10,11,14,18,31,34,35,
+,36,37,38,39,40). Attention is called to the variations in d and A within isomorphous
silicate series (olivines, pyroxenes) in which the Fe/Mg ratio varies. Many
silicate minerals show two or more coupled Mossbauer doublets, ellnlnating from
consideration nany points near the coal point on the plots. Biotite gives an
apparent 2-line spectrum with d and A similar to the coal spectrum, but with
,I decidedly unequal intensities (our measuement), which persist in powdered samples
'
(6), and which can be resolved into a coupled doublet (g), whereas the coal
/ spectrum is symmetrical. Even considering the compositional variations and the
uncertainties of measurement, it is apparent that the non-pyrite iron in coal is
r' distinct from any compound whose Mossbauer spectrum is known. Minerals excluded
>, include oxides and oxyhydroxides, sulfides and related compounds, the carbonates
I
siderite and ankerite, titanian minerals, and the many silicates examined.
4'
Some ferrous complexes of pyridines (35) and 1,lO-phenanthrolines (37) have
similar thmgh non-matching d and A values. These cmparisons suggest that the
non-pyrite iron could be bound to heterocyclic nitrogen aromatic groups in the
coal macemls, or possibly in a clay-like silicate mineral or gel.
f
,
. An attempt was made to correlate the Mzssbauer absorption intensities in the
coals wlth the analytical data. M'dssbsuer spectra were determined on mixtures of
pyrite in carbon, and the fractional peak absorption was plotted awnst the mas8
fraction of pyrite iron. In a few coal samples the M'dssbauer absorption and the
1; analytically determined amount of iron as "pyrite" did agree with the above plot,
but in others the agreement was poor. This may be due to the inadequscy of the
I chemical method for determining pyrite iron, and to differing mrtrix effects in
the coah and the standards. Likevise, a poor correlapn WBB found between the
non-pyrite iron absorption in coals and the amount of organic'' iron deduced fran
the chemical analyses, although the sample with the highest 'organic" iron,
Jewell Valley coal F(a), showed the strongest non-pyrite Morrsbauer spectrum.
i

Table 4. Room-?;ergerature isomer shifts and quadrupole splittings for
po1ycr;stalline iron compounds and n;inerals
This Work Gt her Work
Compound or hlineral d*> hS d+; AS Ref. (
Pyrite, FeS2
Marcasite, FeS,
Siderite, (Fe,Mg)Ch
Ankerite , ( Ca, Mg , Fe )COS
Olivine, (Mg,Fe)zSiOr
Biotite, K(Mg,Feh(AlSbOlo)(OH)2
Tomline, Na(Mg,Fe )3Als(BO3 13 (Sieole) (OH14
Ferrous sulfate, FeSO4'7HzO
Ferrous acetate, Fe( C2H302 )Z
Ferrous stearate, Fe(C1&502)2
Non-pyrite iron in coal
0.54
0.51
1.47
1.46
1.39
1.36
1.30
1.56
1.45
o.66t
1.38
0.60 0'- 55
0.50 0.52
1.78 1.47
1.50 -
2 -95 1.39
2.50
1.27
1.46
Fe+2 1.19
2.38 Fe" 1.44
Fe*3 0.99
3 -05 1-33
2 -23 -
0.70t -
2.62 -
0.61
0.51
1.80
-
3.04
2.41
2.81
2.10
2.61
0.91
* Respect to center of sodium nitroprusside spectrum. ,
t A t liquid nitrogen temperature; no resonant absorption observed at room temperature. ]
9 In m s-'. \.
3
3 e 1 9
-
-
4
i

,'
I
t
f
I
i
I
1
$-
i
h
4
m
3
1
d = ISOMER SHIFT (mm s-')
aigure 2. Representati.re plot of quadrupole splitting versus isomer shift
(relati-re to sodium nitroprusside) for: - ferric compounds (A)
and ninerals (A); ----- ferrous compounds (0) and minerals (0);
ferrate compounds (x), and the non-pyrite iron found in coal (*).
+1
i2

I' 1.3
. /o
9
IO
9
6
a
425
F
F
33 I
3
*9 1 - 1 a . a ,
1'. 0 1'. 1 1'. 2 1'- 3- 1'. 4 1'.5 1:6
d = ISOMER SHIFT (mm S-l)
Figure 3. Enlarged section of Figure 2, showing quadrupole splitting versus isomer shift
relative to sodium nitroprusside for two-line ferrous spectra. Lines connect pairs of
points which are observed together in the same mineral or compound. Dsshed lines indicate'
that a ferric spectrum is associated with the plotted ferrous point. Synthetic compounds:
1-ferrianite (synthetic mica), 2-synthetic pyroxenes, 3-synthetic olivines, 4-welding glass
>-germanium spinel, 6-oxy-salts, 7-carboxylates, 8-halides, 9-pyridine complexes, 10- \
phenanthroline complexes, ll-ferrous hemoglobin. Natural mterials: A-sctinollte, B-
biotite, C-clay minerals, F-fayalite and olivine, M-cllnopyroxenes, P-orthopyroxenes, \
R-ankerite, S-siderite; T-tourmaline, X-tektites, *-non-pyrite iron in coal.
'I
8
6
6
&
8

CONCLUSION
429
In sumnary, several coal sanples contain a form of iron which exhibits a
hitherto unobserved M'dssbauer spectrum, in addition to the well-known pyrite
spectm. Tne non-pyrite iron is ferrous, in a high-spin configuration, in
cctahedral s;metry, and apparently in a rather well-defined state Comparisons
with known compounds suggest that this iron my be bound in the cdl macerals to
heterocyclic nitrocen aromatic groups, though a clay-like silicate mineral or gel
is a possibility. iicssbauer spectrometry can at the very least indicate the
most suitable coal samples for further studies of this iron by chemical and other
metbds.
coal with synthetic ccapounds or analogs. We intend to pursue this line Of
inrestiGtion.
ACKNOWLEKmNTS
It should ultimately permit unamoimous identification of the iron in
'vle wish to ttank Drs. Maurice Deul and Bryan C. Parks of the Bureau of Mines
for their assistance in obtaining some of the coal samples. We are especifrllly
indebted to Mr. Michael Toia for the construction and msintenance of the Mossbauer
spectrometer.
The work at Carnegie,Institute of Technology was performed under
the auspices of tke U.S. Atomic Energy Commission under contract AT(30-1)-844;
Report No. NYO-&44-63.

Brooks J.D. and Sternhell S. (1957) Brown coals. I. Oxygen containing
functional Groups in Victorian brarn coals. Australian J. A S . g. 5
206-221.
Brooks J.D., Durie R.A. and Sternhell S. (19%) Chemistry of brown coals.
11. Infrared spectroscopic studies. Australia? J_. A B . E. 1% 63-80.
Elofson R.M. (1957) Infrared spectra of humic acids and related materials.
Can. J. G. 3 3 926-931.
Dutcher R.R., White E.W. and SpaclmBn W. (1964) Elemental ash distribution
in coal components--Use of the electron probe. Proc. 22nd Ironmakin
Conf., Iron and Steel Div., Metallurgical E. % n I . M X E d .

-5) Sprenkel-Se@ E.L. and Hanna S.S. (l'961t) M'lssbauer analysis of iron in
stone xeteorites. Geochin. Cosnochim. Acta 2 ~ 1313-1931.
,
-7) Gerard A. and Dehelle M. (19s~) Effet M'dssbauer et caractere ionique
des atones de Fer dans les metkorites Orgueil et Cold Bokkweld. Compt.
Rend. Paris a, 17;6-17:.3.
--
.E) Fluck E., Kerler W. and Neuwirtil Vf. (1963) The M'dssbauer effect and its
sipificance in chemistry. Anhew. Chen. Internat. EL. 5 277-22{.
-3) Wertheim G.K. (1964) M'dssbauer Effect, Principles and Applications.
Academic Press, Inc., New York.
IC) Goldanskii V.I. (1964) M&sbauer Effect and its Applications
Chemistri. Consiltants Bureau, New York.
'1) Herber R.H. (195;) Introduction to M'dssbauer spectroscopy. J. Chem. Educ.
fil, 130-1C7.
J2) DeBenedetti S., Barros F. des. and Hoy G.R. (1966) Chemical and structural
effects on nuclear radiations. A=. E. E. g. & 31-82.
'3) Herber R.H. and Hayter R.G. (19&) Mgssbauer effect in, cis-trans isomers.
J.. &. Chem. SOC. g, 301302.
'4) Spijkemn J.J., &egg F.C. and DeVoe J.R. (1965) Standardization of the
differential chemical shift for FeS7, M'dssbauer Effect Methodology,
pp. 113-l2@. Plenum Press, New York.
j5) Thiessen C. (1945) Chemistry of Cos1 Utilization (Ed. H.H. Lowery),
Vol. 1. John Wiley, New York.
36) Mitra G.B. (195h) Identification of inormic impurities in coal by the
X-ray diffraction method. Fuel 3& 316-330.
'7) Ecknause M., Harris R.J., Shuler W.B. and Welsh R.E. (1966) Measurement
of the lifetime of the 14.4-keV level of 57Fe. Proc. Phys. e. London
a 187-1i36.
~ 8 Obenshain ) F.E., Roberts L.D., Colemn C.F., Forester D.W. and Thomson
J.O. (1965) Hyperfine structure of the 14.4-keV y-ray of s7Fe in hydrated
ferric smmonium sulfate as a function of the magnetization of the salt.
~2. E. Lett. & 36j-369.
'9) Wignall J.W.G. (1966) M'dssbauer line broadening in trivalent iron compounds.
5. =. P*. 4k, 2462-2467.
iC) Margulies S. and Ehrrcan J.R. (1961) "ransmlssion and line brcadening of
resonance radiation incident on a resonance absorber. Nucl. Instr. Meth.
131-137-
j1) Epstein L.M. (1962) MGssbauer spectra of some iron complexes. 5. E.
P2. 2731-2737-
1

(52) Gib3 T.C. and Greenwood N.N. (19%) The Fidssbauer spectm of the tetra-
chloroferrate(I1) ion. >.
(j;) Lang G. and Marshall W. (1966) M'dssbauer effect in some haerno~lo3in
corzpounds. Proc. Phys. E. London 87, 3-34.
(j3) t4atr.u- H.B., Sinha A.P.B. and Yagnik C.M. (L465) The M'dssbauer spectra of
spinel oxides containing ferrous ion. S a
State C=. & 401-403.
(4G) Frank E. and Abeledo C.R. (1566) Mzssbauer effect in iron (111)
dithiocar3amates. Inorg. z. & 1433-145>.

1133
REACTIONS OF COAL AND RElATED MATERIALS IN MICRWAVE DISCHARGES IN H2, H20 AND Ar
Yuan C. Fu and Bernard D. Blaustein,
U. S. Bureau of Mines, Pittsburgh Coal Research Center, .
4800 Forbes Avenue, Pittsburgh, Pennsylvania 15213
INTRODUCTION I
Recent investigations on the reaction of carbon in a high frequency discharge have
shown that it produces methane, acetylene and minor amounts of other hydrocarbons
in the hydrogen dischar e,=/ but produces primarily an H2-CO mixture downstream
from a water discharge.-/ 9 Similar vork with respect to coal, however, is almost
non-existent except that carried out by Letort et a1.4/ and by Pinchin.'/ Though
some work on coal in a plasma jer6-91 has been reported, the characteristic of
those reactions is generally the thermal decomposition of coal by rapid heating
to high temperatures in an inert atmosphere.
In a microwave-generated discharge, hydrogen or water vapor can be excited or
dissociated into atoms, ions and electrons at temperatures much lower than those
attained in a plasm jet. The present study is concerned with the reactions of
various coals, a polynuclear hydrocarbon, and graphite, in microwave discharges
in H2, water vapor and Ar. The results are compared in terms of the product yield
and distribution in each type of discharge. Differences observed between the
reactions of the various coals and the other materials suggest that the amount and
type of volatile material, carbon content, and the type of the carbonaceous material,
as well as the type of the gas discharge, are all factors affecting the product yield
and distribution. The difference between coal and graphite - . in their behavior toward
water vapor in the microwave discharge is of particular interest.
discharge, the graphite yields almost no hydrocarbons but only an H2 + CO mixture
while the coal yields appreciable amounts of C2H2 and CH4 in addition to H2 + CO.
Experiments using DzO and Ar.discharges ,indicate that the D20 and the water actually
do participate in the reactions with coal to form hydrocarbons. It is also demon-
strated that the C2H2 yield from coal in the hydrogen discharge can be drastically
increased by condensing at a low temperature part of the primary products formed
during the discharge reaction.
ACKNOWLEDGEMENTS
1 ,
In the water
The authors wish to thank Dr. Irving Wender for his valuable discussions, Gus
Pantages for his technical assistance, A. C. Sharkey, Jr. and Janet Shultz for
their mass spectrometric analyses, and John Queiser for his infrared analysis.
EXPERIMENTAL
The experiments were carried out in a static system 4th the discharge produced by
a Raytheon microwave generator in the air-cooled Ophthos coaxial cavity at 2450Mcl
sec. Chemical analyses and origins of the vitrains of the different coals used are
given in table 1. All the vitrains were -200 mesh. Ultra purity spectroscopic
graphite powder (325 mesh, United Carbon) and chrysene (c18H12) were used for
comparison. As reactant gases, a 9.7:1 H2-Ar mixture, H20-Ar mixtures and Ar were
used. The H2 and At were obtained from cylinders and passed through a liquid N2
trap prior to storage. The uater vapor was obtained from distilled water degassed
in high vacuum.
In the experimental procedure, a knom weight of graphite or coal powder was placed
in a cylindrical Vycor tube (ID - 11 m, vol - 32 ml) and degassed in a high vacuum
at an elevated temperature-(150°C for graphite and 100°C for coal) for a half to
one hour to remwe the moisture and the air adsorbed on the carbon. ' A known
pressure of the reactant gas or gas mixture measured by a Pace Engineering pressure

434
transducer vas then introduced into the reactor tube. For the H20-Ar mixture, a
known pressure of the water vapor was first introduced to the reactor containing
the degassed carbon, and then a known pressure of the argon was introduced while
the water was condensed in the end of the tube by dry ice and acetone. The portion
of the tube cooled by the dry ice was so small compared to the total volume that no
corr=ctim CI? preee~re reeding spp~~r! to he necensary. The dinrhnrgc wan then
produced with the carbon in the discharge zone. The gaseous products were analyzed
by -8s spectrometer.
1
'Lhe solid product obtained from the high volatile bituminous coal in the H2 discharge
was analyzed by infrared spectrometer. The sample was obtained either by scraping
the reactor wall or by solvent (benzene or acetone) extraction. In some instances, a
reactor consisting of a tube divided by a fritted Vycor disc was used. The coal vas
placed on the disc and KBr powder was introduced on the other side of the disc.
During the discharge, the lovcr end of the tube was immersed in liquid N2, thus
permitting the condensable low volatile products to pass through the disc end be
adsorbed on the surface of the KBr powder. This was later removed, pressed into
a pellet, and examined by infrared analysis.
TABU 1. - Analysis of vitrainr (moisture free basis, percent)
0 Vola ti le I\
C H N S (by diff.) mtter
Anthracitdl
Low volatile
91.06 2.49 0.96 0.83 2.89 1.77 6.1
pninou&l 89.57 4.67
4 . ,,&k~$~IA
81.77 5.56
1.25
1.71
.81
.97
2.17
5.87
1.53
2.06
20.2
39.2
9
. -5/ Lignite21 66.45 5.40 .31 1.40 22.84 3.06 44.0
2/
3/
4'
Dotrance nine, &high Valley Coal Co., Luzerne County, Pennsylvania.
Pocahontar lo. 3 Bed, Buckeye No. 3 Mine, Page Coal and Coke Co.,
Stephenson, Wyoming County, West Virginia..
Bruceton, Pittsburgh Bed, Allegheny County, Pennsylvania.
Beulah-Zap Bed, North Unit, Bculah Mine, Knife liver Coal Ubi- Co.,
Beulah, Mercer County, North Dakota.
RESULTS AND DISCUSSIOH
All result. were obtained from experiments using 5 mg of the carbonaceour arterial
in each discharge at 35 watts of power input. For consistency, an initial total
pressure of about 25 UD y1s used for most runs, but in the runs with H20 vapor it
tar varied in the range of 12 to 25 m.
The time of the discharge reaction -8
intended to be 60 seconds in mort NIIl, but the dirchrge would 80t rustaio itrelf
in s oy runs. It ir rurpected that the rolid product (probably polymuclur hydro-
carbons) formed u y teed to draw away the electroar in the dircharge to form negative
ions and create a rhortage of energetic species in the gas phase. &never, under the
experimental condition. crployed, a large part of the reaction occurred within 30-60
seconds, and prolaged treatment resulted .ply in IQY chnge in the product dirtri-
bution and a little lacreare in the extent of the garification. A few rum were
ude for ar long .E 3 rinutes.
Hydrogen-Argon and Argon Dirchrges
In prelirinary runr with graphite in an 82 dirchrge, we attempted to interpret
the data on the baris of the hydrogen balance before and after reaction. It
appear., horcver, that appreciable w n t r of hydrogen were consumed in the
fotration of rolid product or were taken up by the carbon. Therefore, to allov

435
interpretation of the data, Ar was introduced into the reactant gas as an internal
standard. The presence of Ar in the system gives no noticeable effect on product
type or distribution.
Table 2 summarizes the results obtained from the reactions of each material in
microwave discharges in an H2-Ar (9.7:l) mixture and in Ar. In the Ar discharge,
coal is partially gasified to give Hq, CO, and C2H2 as the major gaseous products,
CQ2, CHq, and minor amounts of other hydrocarbons (C2 and C3 hydrocarbons.
biacetylene and benzene were detected), in addition to a solid prodkt and residual
char. In the H2-Ar discharge, the percent of carbon forming gaseous hydrocarbons is
increased as compared with the Ar discharge; the total amount of carbon gasified is
also increased. In this discharge, C2H2, CO, and CH4 are the main constituents of
the gaseous product, and minor amounts of other hydrocarbons are also formed as in
the Ar discharge. Comparison of the several 60-second runs indicated that the net
amount of hydrogen remaining after the reaction was decreased, except for high
volatile bituminous coal and lignite. Besides that converted to hydrocarbons, part
of the hydrogen was probably consumed in the formation of solid product or was
taken up by the coal residue. With high volatile bituminous coal and lignite,
gasification was extensive and resulted in a net increase of hydrogen.
The percent of carbon converted to gaseous hydrocarbons increases with the volatile
matter of coal as shorn in figure 1, suggesting that a rapid release of the volatile
matter and its subsequent gas phase reaction in a discharge may be the determining
factors. Figure 1 includes the data for graphite. Lignite, with the highest
volatile matter, however, gave only small amounts of hydrocarbons in both discharges.
The high oxygen content of the lignite results in higher yields of carbon oxides and
water but apparently inhibits hydrocarbon formation. Water vapor, in a discharge,
can reverse the reactions of hydrocarbon fomtion by reacting with the hydrocarbon
species to form carbon oxides and hydrogen.-
Chrysene did not behave particularly different from coal in both discharges; the
extent of gasification was appreciably small, approximately the same as that for
anthracite. Interestingly, the carbon contents of the chrysene and the anthracite
are also close to each other and are much higher than the other coals.
For coal in general, the higher the carbon content the lower the volatile matter.
The extent of the reaction of graphite with H2 was very much smaller under
comparable conditions, probably due to the absence of volatile matter. The main
products from graphite were C% and C2H2; however, the hydrogen balance Indicated,
for example, that with an initial pressure of the H2-Ar mixture of 25 m Hg, the
percentage of the initial hydrogen present in each component of the product is H2,
77.6; CH4, 9.2; C2H2, 3.3; C2 + C3 hydrocarbons, 1.6; and the remainder (8.3%) was
apparently consumed in the formation of polymer or was taken up by the graphite.21
It is interesting to note that C2H2 accounted for 75-92% of the gaseous hydrocarbons
produced from coal in both discharges. Hydrogasification or carbonization
of coal usually gives CHq as the major hydrocarbons. It has been suggested that, in
a microwave hydrogen discharge system, transport of carbon from the solid to the
vapor phase takes place by bombardment of the carbon by energetic ions and electrons,
and that the ’pr”seous carbon species could react with H atoms or CH species to form
hydrocarbons . 2 1 A sfmilar mechanism appears co apply to the system containing
coal except that H and CH species would also be evolved from the coal in addition
to gaseous carbon species, even in the absence of an initial hydrogen. The
predominance of C2R2, however, is also observed in rapid heating of coal by various
high temperature methods such as plasma jet,=/ laser be&/ and flash heating.Ef
etc. A cornpariron of our experimental results with those obtained vlth an Ar plasma
jet by Bond et al.l/ indicates that the percentage of carbon converted to C2H2 In
our system is somewhat lower, but the effect of volatile matter on the hydrocarbon
4

436
yield is quite similar. The conversion of carbon to C2H2 for the high volatile
bituminous coal (VM 5 39.2%) in table 2 is 9.2% in the H2-Ar discharge as compared
to 12.5'1. in the Ar plasma for a coal of similar volatile matter content.
TABLE 2. - Reactions of coal and related materials
in microwave discharges of H? and Ar 1
Pressure,
Percent of c present as ,/
Uaterial
m
H,, Ar
Time, Yield x lo4, mole/g of solid
sec H? clll C2H2 CO CO?
Gasebus Gaseous
products hydrocarbons
hvab 22.7 2.3 60 63.9 4.7 31.1 21.1
22.7 2.3 60 53.6 10.2 28.6 20.2
23.4 2.4 60 12.5 4.5 27.0 23.6
22.8 2.4 90 11 3,9 29.8 22.2
21.8 2.3 105 L/ 4.8 21.0 ' 16.9
21.9 2.3 110 11 3.6 20.0 16.0
- 25.7 17 3.3 .1 1.2 7.1
- 24.0 60 56.4 .9 13.4 23.0
- 23.6 180 44.0 1.1 16.5 22.6
lvb 23.8 2.5 34 1/ 4.1 15.5 7.3
21.6 2.2 60 1/ 3.8 17.5 8.1
- 23.5 60 15.3 .3 3.6 7.7
Lignite 21.6 2.2 60 25.4 2.1 9.3 66.2
22.6 2.3 180 22.9 2.1 7.7 80.3
20.7 2.1 180 26.0 2.1 7.3 69.2
- 24.7 32 50.4 .7 6.9 61.7
- 23.0 60 18.4 .6 5.0 68.4
Anthracite 21.6 2.2 60 11 4.2 8.3 4.0
22.6 2.4 60 11 4.6 9.6 3.0
- 24.0 60 .3 trace trace 2.0
Chrysene 22.7 2.4 10 11 6.5 12.8 2.0
20.0 2.0 30 11 10.4 5.8 trace
- 22.0 30 lz.6 .8 2.6 trace
Graphite 22.4 2.3 60 11 3.5 2.5 trace
22.6 2.4 60 11 3.4 1.8 trace
11 Net decrease of hydrogen was indicated.
0.2
.2
.4
.2
.2
.2
.1
.2
.2
.1
.1
trace
3.5
1.8
1.7
3.2
3.2
trace
trace
.1
trace
trace
trace
trace
trace
13.6
13.1
12.8
13.1
9.9
9.4
1.5
7.7
8.7
6.1
6.7
2.1
16.7
18.3 '
16.1
14.1
15.1
3.6
3.9
.3
4.6
3.4
.9
1.2
1.0
I
10.4
10.3 \
9.4
9.9 4
7.4
7.0
-4 \
4.3
5.3 '
5.1 \
5.6
1.1
4.1
3.3
::: /
3.1
3.4 ~
4.4
3.4 \
.9 I
1.2
1.0 \
Solid Product I
The solid product obtained from coal is brownish and is similar to that usually
observed from the thermal treatment of coal. Uo appreciable amount of solid product
was formed from anthracite or from lignite. Though the extent of the gasification
for the chrysene was not appreciable, the original white powder was instantaneously
converted to a broun solid upon the initiation of both discharges. The infrared <
spectra of the solid product and the residual char obtained from the high volatile
bituminous coal in the H2-Ar discharge showed the usual aliphatic C-R bands and
some weak aromatic bands which are typical of pitch and coal.
3
* 7
1
Effect of Cooling by Liquid Nitrogen
Since the indications are that the wnter formed can retard the hydrocarbon formation
and that the hydrocarbons produced may undergo further destructive reactions, it can
be expected that a rapid quenching of the primary products should give a pronounced
effect on the result. Experiments were carried out in a reactor consisting of the
i t

437
tube divided by a fritted Vycor disc. The coal was placed on the disc, and was
subjected to reaction in the H2 discharge while the lower end of the tube was cooled
in liquid N2. Though the process of condensing water and some hydrocarbons at this
temperature is diffusion controlled, the effect is pronounced, a's shown in table 3.
For both the bituminous coal and the lignite, the yield of the hydrocarbons, CqH2
in particular, was greatly increased. The amount of Hg remaining and the amount of
CO produced after the reaction were also decreased. Therefore, the increase of
hydrocarbon yield can be attributed mainly to subsequent hydrocarbon formation by
reaction of H2 and co; this hydrocarbon yield is greatly enhanced b$ rapid removal
of H20 formed.
Water-Argon Discharge
There is a marked difference between the products obtained from graphite and coal in
a water discharge. In the discharge in H20-Ar mixtures, as shown in table 4, graphite
yields hydrogen and CO but practically no hydrocarbons; while the coals yield an
appreciable amount of CZHZ and some CH4 in addition to H2 and CO. The amounts of
Hq and carbon oxides produced in the reaction with graphite rere stoichiometric.
The extent of gasification was also rpuch greater for the coals than for the graphite.
The active hydrogen species produced in the water discharge did not react further
with the graphite or the CO formed from it to produce significant amounts of hydro-
carbons. This is further evidence demonstrating that the presence of water vapor
retards hydrocarbon formation.
For the reaction of a given coal in aR2aAr discharge, an initial H20 pressure of
less than 12 mm Hg appears to give the optimum gasification end hydrocarbon
production. Higher initial H20 pressures cause a decrease in both gasification
and hydrocarbon production. So long as the H20 pressure is not at its highest
values, more hydrocarbons are formed than in the Ar discharge. (Also, with the
higher initial Hfl pressures, the discharge could not be initiated readily and
would not sustain itself for as long as 60 seconds, perhaps due to too large an
increase in the total gas pressure in the reactor.) The data seem to indicate
that 60 seconds may have been too long a period for the maximum production of
hydrocarbons. The formation of the hydrocarbons should be a maximum at the time
when a plateau of the extent of coal gasification is attained, and prolonged
treatment probably allows the remaining H20 vapor to diffuse into the discharge
zone, giving an adverse effect. It was also noticed that, at an initial H20
pressure of less than 12 mm, the hydrogen content of the products exceeded that
which could possibly be derived from the stoichiometric amount of the H20
initially present.
These results seem to indicate the following. The active hydrogen species formed
in the Hq0 discharge participate in the reactions which lead to the formation of
hydrocarbons from (some of) the species derived from the coal. This occurs
despite the retarding effect of H20 on hydrocarbon formetion. (If the H20
pressure is too high, this latter effect decreases the hydrocarbon yield.)
Presumebly, the coals when gasified supply enough CB species to allow hydro-
carbons to be formed,even though the active oxygen species present react with
some of the gaseous carbon and CH species. On the other hand, graphite produces
negligible amounts of CH species in a H20-Ar discharge, and therefore cannot
produce hydrocarbons since the reaction of the CO formed with the active hydrogen
species is retarded by the H20 present.
Lignite, with its high volatile matter content, was extensively gasified but gave
a relatively low hydrocarbon yield, presumably due to the inhibition by its high
oxygen content. For lignite, the production of COq was also higher in the R20-Ar
discharge than in the Hq-Ar or Ar discharge. Again, the extent of gasification
for chrysene was rather small.

?-!???
0 mv,
Uh\OhIe
. . . .
eirlrl
00000
wwwww
r l r l
I
*
438 ,
I
,

439
Deuterium Oxide-Argon Discharge
It is of interest to obtain further evidence as to whether water is actually involved
in the reaction with coal to form hydrocarbons in addition to the production of
H2 + CO. The gaseous product obtained from the reaction of the high volatile
bituminous coal in a D20-Ar mixture was analyzed by the high resolution mass
spectrometer. At a resolution of 1 part in 20,000, precise masses for doublets and
in some instances triplets that occurred at the same nominal masses could be used
to identify completely deuterated, monodeuterated and nondeuteratedl species present
in the product. For example, peaks due to C2H2 (mass = 26.0156), C2D (26.0141) and
CN (26.0031) were observed at the nominal mass of 26 and peaks due to C2D2 (28.0282),
N2 (28.0061) and CO (27.9931) were observed at the nominal mass of 28. The whole
spectrum up to the nominal mass of 44 contained peaks attributed to all species
present including minor amounts of oxygenated and N-containing compounds, but the
major products were H2, HD, D2, C2H2, C2HD, C2D2 and partially deuterated methanes
in addition to D20, DHO, H20 and carbon oxides. Thus it is apparent that D20 is
initially dissociated into D, OD and possibly an active oxygen species which in
turn recombine to give D20 and D2, or react with the active species derived from
coal such as H, CH, C1, Cg and CO, etc., to give DHO, HD, CO, C02 and deuterated
hydrocarbons.
CONCLUSIONS
In the microwave discharges of H2, H20 and Ar, coal is gasified to give gaseous
hydrocarbons and carbon oxides plus a solid product and residual char, Hydrogen
I is also produced either by dissociation of the water vapor or by devolatilization
of the coal in the H20 and/or At discharges. But, in most cases in the hydrogen
discharge, hydrogen is consumed rather than produced (except for hvab coal and
lignite). The yield of the hydrocarbons is highest in the hydrogen discharge and
that of carbon oxides is highest in the water discharge. Both the hydrogen and
{ the water discharges give greater extents of gasification and yield more hydro-
/ carbon products than the Ar discharge, indicating the occurrence of gas phase
1 reactions of H, OH, and active oxygen species with the active species derived
from the coal. The gasification of coal in the water discharge is of particular
' interest because it produces H2 + CO ( - 1:l) plus C2H2 and CQ. The high oxygen
J content of the lignite results in a higher yield of carbon oxides but apparently
inhibits the hydrocarbon formation in the discharge. The hydrocarbon yield from
the lignite or other coals in the hydrogen discharge, however, can be increased
,' dramatically by rapidly condensing part of the product species produced, especially
i H20, during the discharge reaction.
L
C2H2 accounts for as nuch as 9ZZ of the gaseous hydrocarbons produced, and, excepting
for lignite, the amount of the hydrocarbons is related to the volatile matter of coal.
The extent of nasification. however. is increased with the volatile matter content
of coal including lignite.
Thus, if the hydrocarbons are formed by the recombination
of the species derived from the volatile material and other species present in the
discharge, another intererting study would be to employ a flow system in khich coal
is devolatilized by ordinary means and the evolved gases are subsequently reacted
in a discharge.
REFERENCES
1. Blackwood, J. D. and McTaggart, F. K. Reactions of Carbon with Atomic Gases.
Australian J. Chem., 12, 533 (1959).
1. Shahin, n. n. Reaction of Elementary Carbon and Hydrogen in High-Frequency
Discharge. Nature, 195, 992 (1962).
3. Vastola, F. J., Walker, P. L. Jr., and Wightman, J. P. The Reaction Between
Carbon and the Products of Hydrogen, Oxygen and Water Microwave Discharges.
Carbon, 1, 11 (1963).

4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
440
Letort, M., Boyer, A. F., and Payen, P. Attempting Hydrogenation of Coal with
Atomic Hydrogen. Bull. SOC. Chim. France, 1589 (1963).
Pinchin, F. J. The Reaction of Coals with Atomic Hydrogen. Private
Comnication.
Bond, R. L., Galbraith, I. F., Ladner, W. R., and McConnell, G. I. T.
Produccion of Acecyiene from Cuai, iising a Piasma iec. Nacure, =, 1313
(1963).
Bond, 8. L., Ladnei, W. R., and McConnell, G.
a Plasma Jet. Fuel, 45, 381 (1956).
I. T. Reactions of Coal in
l
Graves, R. D., Kawa, W., and Hiteshue, R. W. Reactions of Coal in a Plasma
Jet. Ind. Eng. Chem., 5.59 (1966).
Kawana, Y., Makino, M., and Kimura, T. Formation of Acetylene from Coal by
Argon Plasma. Kogyo Kagaku Zasshi, 2, 1144 (1966).
Blaustein, B. D. and Pu, Y. C.
Formation of Hydrocarbons from H2 + CO in
Microwave-Generated Electrodeless Discharges. Paper to be presented at 153rd
National ACS Meeting, Miami Beach, Fla., April 1967.
Montet, G. L. Retention of Protons by Graphite. Proc. Fourth Conf. on Carbon,
Pergamon, New York, pp. 159-164 (1960).
Sharkey, A. G., Shultz, J. L., and Friedel, R. A. Gases from Flash and Laser
Irradiation of Coal. Nature, 202, 988 (1964); Coal Science, Advances in
Chemistry Series 55, R. F. Gould, Editor, Am. Chem. SOC., Washington, pp- 643-
649 (1966).
Rau. E. and Seglin, L. Heating of Coal with Light Pulses. Fuel, 2, 147 (1964).
1
i

I2
c
C
01
? IO
e 6
LL
0
z
13
0
0 hvab vitroin
A Ivbvitroin
X Anthrocite vitroin
0 Graphite
0 Lignite vitroin
I
Flgure I -Effect of volotile matter on yield of hydrocorbons in H2-Ar discharge
(60 second run)
I
12-29-66 L-9741

442
CHEMICAL REACTIONS IN A CORONA DISCHARGE I. BENZENE
1
. .. M. W. Ranney'.
Wm. F . O'Connor
Department of Chemistry
Fordham University
New York, New York 10458
INTRODUCTION
Corona discharges (silent) have been used to initiate chemical reactions for
well over 150 years'with much of the early work being confined to inorganic gases.2 However,
the only commercially significant development over' the years has been the process
for synthesizing ozone by subjecting oxygen to a corona discharge3 (ozonizer). Most early
investigators encountered serious difficulties in studying the interaction of high voltage
electricity with organic molecules. Equipment was unreliable and dangerous, and the
complexity of the reaction product mass precluded definitive performance evaluation. Slnce
World War 11, technology has advanced to the point where high frequency power can be
generated and controlled at reasonable cost and many new dielectric materials such as
fused quartz, alumina and mica mat are a~ailable.~ Modern analytical techniques afford
the opportunity to determine the composition of the product mix.
Corona electrons are accelerated by the applied voltage to an energy.levelof 1
10-20 electron volts, which is sufficient to break the covalent bond. This represents a very
efficient approach when compared with high energy radiolysis (m.e.v. range) where the
actual chemical work is effected by secondary electrons with an energy of 10-25 e.v.,
formed after a series of energy-dissipating steps. Potentially, corona energy can deliver
electrons to the reaction site at the desired energy level to give products which are not
readily obtainable by more conventional means. The energy available in the corona dis-
charge is somewhat above that commonly encountered in photochemistry (up to 6 e.v. ).
Bertholot reduced benzene to c@8 in 1876 in what appears to be the first
exposure of an aromatic compound to the corona discharge. Losanitsch obtained the
following products from benzene in an electrical discharge: ( CgH6)2, biphenyl, (C72Hgg)
and (C6H6)90,6 Benzene vapor treated at 300° in an ozonizer tube gave resinous products
with a 6/4 carbon-hydrogen ratio.7 Linder and Davis exposed benzene vapor to 37,000
volts and found biphenyl, gaseous products and evidence of polymerization.8 The early
work yith benzene reactions in various ty s of electrical discharges is reviewed in
considerable detail by Clocker and Lind. 'Brown and Rippere investigated the hydrogenation
of benzene flowing down the walls of,an ozonizer tube and found 1,3-and 1,4-
cyclohexadiene, biphenyl and a resinous mass which gave an infrared spectrum con- i'
sistent with polystyrene. !4 (
In recent years, benzene has been subjected to direct electrode discharge
10 1
and glow discharges induced by microwave'' and radiofrequency'l energy. In this paper, I
we report some of our observations for the reaction of benzene in a corona discharge.
I
'7
1
/
!
'I
1
)
>
i
1,
\d t
!'
\

443
A PPA RA TUS
The corona reactor system used in this study is shown in Figure 1 and the reactor
/details are given in Figure 2. The system is designed to run at atmospheric or reduced pres-
'*sure and, with slight modification, under recycle conditions. The use of the threaded rod
;center electrode affords a more uniform and higher treating potential than is obtained with a
'cylindrical electrode. l3 The electrode threads concentrate surface irregularities thereby
I
'eliminating severe discharge points. The copper electrode serves as a cooling coil and
affords direct observation of the corona. In a typical run, the benzene reservoir temper-
-ature is set at 53O, the helium is adjusted to 100 ml/min. and after a two minute purge, an
electrical potential of 15,000 volts is applied. A brilliant blue corona is established, the
intensity being a.function of helium flow, pressure, benzene content and the applied voltage.
!Benzene traverses the corona reactor as.a vapor and is condensed by the cold water trap
nd collected. Yellow solids gradually deposit in the flask at the bottom of the reactor and ,
's eventually adhere to the dielectric surfaces in the reactor. Some benzene and the more
volatile reaction products are condensed in the -70 and -195" traps. See experimental
section for further details.
,(
RESULTS AND DISCUSSION
t . J
' .
The excitation of benzene vapor in a corona discharge provides an 8.5% conversion
to identifiable products. The major products are a benzene soluble polymer (6%):
la benzene insoluble polymer (0.7%), biphenyl (0.3%), ,and acetylene (1%). The composition
of the product mix obtained is given in Table 1. The overall yield obtained under
'our experimental conditions is similar to that produced in a radiofrequency glow discharge
12
I;( lo%, with longer residence time) and somewhat higher than Streitwieser obtained using a
microwave'' induced glow discharge (5%). The corona discharge gives a considerably
I higher proportion of polymer than is obtained from these other energy sources. Benzene
ubstitution products such as toluene, phenylacetylene and ethylbenzene were observed in
he microwave discharge but were not noted in this study or in the radiofrequency invesigation.
The high yield of fulvene in the radiofrequency discharge is surprising in view of
ts tendency to polymerize or add oxygen under rather mild conditions. The sum of the
arbon and hydrogen analyses for the polymeric fraction obtained by Stille in the glow disharge
of benzene (86-94 per cent) suggests some fixation of oxygen and/or nitrogen during
ischarge or product work-up. l2 The pressure for these electrodeless glow discharge
tudies was of the order of 20 mm. or less, while the corona reactor was operated at
tmospheric pressure. 'Schiiler, using an electrode discharge in benzene, obtained proucts
similar to those produced in the microwave glow discharge. lo
1
The reaction products obtained in this study are roughly categorized as the
cnzene fraction, biphenyl fraction and polymeric material for purposes of discussion.
,fter 'initial studies indicated that the components of the -7O.and -195O traps were
imilar to those in the benzene trap, these samples were combined. The acetylene
ontent of the low temperature traps was determined by weight loss prior to mixing
iith the benzene.
cnzene Fraction
The exposed benzene, as collected in the cold traps, is bright yellow. The
allowing reaction products have been identified in this fractlon: I, 3-cyclohexadiene,
,4-cyclohexadiene, fulvene and cyclohexene. The approximate per cent yield for each
f these consntuents is given in Table 2. The data were determined using a 100 foot

helium
444
CORONA REACTOR SYSTEM
Figure I
condenser 400 mm
threaded steel rod
electrode 5/16"x 36"
DETAIL OF CORONA REACTOR
Figure 2
7740 Boroslltcote gloss
dielectric, llmm OD
thermocouple well
,1/4" copper tubing electrode
and cooling coil
Gap-4mm
Internal volume -164 ml
thermocouple well
24/40 ground gloas joinf
10 wet
I
\
\
'1
%
I
1
F

Product
445
TABLE 1
Product Dstribution in the Corona Discharge of Benzene, 15KV
Benzene insoluble polymer 0.7
Benzene soluble polymer (a) 6.0
Biphenyl 0.3
0-. m-, and p-Terphenyl
Phenyl, benzyl substituted
0.1
cyclopentenes (b) 0.3
1,3-and 1,4-Cyclohexadiene 0.1
Fulvene 0.01
Acetylene 10
TOTAL 8.5
Yield-Per Cent Yield-Per Cent Radiofrequency@)
Benzene Exposed of Product Mix Yield, Per Cent
-
8.2
70.5
3.5
1.2
3.5
1.2
0.1
11.8
100.0
2.0
N.D.
N.D.
N.D.
I,O
2.0 (f)
10.0
(a) Polymer fraction includes materials from 250 molecular weight and up.
(b) Includes several phenyl and benzyl substituted cyclopentenes similar to biphenyl and
o-terphenyl in vapor pressure. See discussion.
(c) See reference No. 12.
(d) N. D. -- not determined.
(e) Benzene and toluene soluble polymer.
(f) Acetylene l%, allene 1%.
support-coated open tubular column with a squalane liquid phase and appropriate calibration
standards (Figure 3). No significant change is noted in the cyclohexane content when
compared with starting benzene. The increase in the methyl cyclopentane peak is ascribed
'to its higher vapor pressure thus affording some accumulation of this material in the
benzene distillate and a concomitant depletion in the benzene feed stock. Cyclohexene is
produced in very low yield, as expected, because of the degree of hydrogenation required
during its short residence time in this single pass reactor.
The reduction of benzene to the cyclohexadienes is interesting as
1,3-cyclohexadiene is therm'odynamically unstable with respect to benzene, cyclohexene
and cyclohexane. l4 The synthesis of 1,3-cyclohexadiene under these conditions is
analogous to the conversion of oxygen to ozone which also represents an energetically
unfavorable process. The momentary energy input and the rapid removal of the activated
species from the reaction zone affords this material. in low yield., As expected, the
more stable 1,4-cyclohexadiene is produced in higher yeld (3/1). Additionally,
1,3-cyclohexadiene would be expected to polymerize at a rapid rate under these conditions.
The thermodynamic instability of 1,3-cyclohexadiene is demonstrated by its
gradual disappearance from the sample after a few days standing at room temperature.
The yields of both the l,4-and 1,3-cyclohexadiene are somewhat higher than previously
reported by Brown and Rippere. The yields reported by these workers after
24 hours exposure to a 15 KV'corona discharge in a counter-current hydrogen stream
were 0.02 and 0.01 per cent respectively. The major variable in technique which likely
accounts for this difference.in yields is the use of benzene vapor in our work, versus
the hydrogenation of a thin liquid benzene film running down the annular dielectric sur-
.
face by Brown and Rippere.

446
The ultraviolet spectrum. for the benzene fraction was obtained by running
differentially against pure benzene in isooctane. A broad absorption band with a maximum
at 258 rnp is ascribed to the 1,3-cyclohexadiene. An additional absorption at 242 rryl and
a tailing into the visible region with a broad maximum at 360-370 r y is also observed.
These spectral features are consistent with those reported for fulvene, the non-aromatic
isomer of benzene.15 Blair and Bryce-Smith found fulvene in the photochemical irradiation
of benzene in what appears to be the first direct conversion of an aromatic hydrocarbon to
a non-aromatic hydrocarbon. l6 Fulvene co-distilled with benzene, but could be separated
from benzene using the 100 ft. squalane column (Figure 3). Fulvene prepared by the
method of Meuche gave the same retention time and spectral features.I7 Additional evidence
for assignment of the fulvene peak included its disappearance from the chromatogram after
the sample was refluxed with maleic anhydride (color disappears ) or exposure to a free
radical catalyst. The Diels-Alder reaction product with maleic anhydride was isolated and
hydrolyzed to give the adduct, 7-methylene-5-norbornene -2, 3-dicarboxylic acid. The
melting point and infrared data for this adduct were identical with the sample prepared
using fulvene synthesized from cyclopentadiene and formaldehyde. l2
The formation of the 1,3,5-hexatrienyl diradical has been suggested as an
intermediate in the photochemical decomposition'' and the radiofrequency discharge12 of
benzene. Material isolated in the photochemical excitation gav.e a W spectrum similar
to 1,3,5-hexatriene, but displaced by 7.5 mp and, more disturbing, the investigators
were able to separate the material from benzene by fractional distillation. No experi-
mental evidence was given in the radiofrequency study. In our work, 1,3,5-hexatriene
could not be detected (less than 100 ppm) in the benzene fraction using the squalane
column. No spectral evidence was noted in the ultraviolet although the region of in-
terest is complicated by the presence of 1,3-cyclohexadiene. l9
The unidentified peak in the chromatogram (Figure 3) gives a negative
test for an acetylenic hydrogen (ammoniacal cuprous chloride solution) and does not
disappear after the sample is subjected to free radical catalysis for several hours. Thus,
the open chain, acetylenic isomers of benzene, hexa-I, 3-diene-5-yne. 1,4-hexadiyne,
1,s-hexadiyne and l.S-hexadien-3-yne do not appear to account for this peak. The
possibility of valence isomers of benzene such as bicyclo (2.2.0) hexa-2, 5-diene,
"Dewar benzene", were excluded based on the observation that the peak was stable to
prolonged heating at looo. The valence isomers would be expected to convert to
benzene under such conditions. 2o
TABLE 2
(a)
Product Distribution in Dscharged Benzene Fraction
Product Yield % Benzene Exposed
1.3 -Cyclohexadiene 0.03
1.4 -Cyclohexadiene 0.09
Fulvene 0 01
Cyclohexene 0.01
Unidentified 0.02
(a) Quantitative data obtained by GLC using squalane column and appropriate calibration
standards. I
I li
\
\

I
f I
I
" I
I
-
447
~.
Figure 3
Analysis of Benzene Fraction Column
I00 f t a 0020 In ID Support-coated open
Tubular with Squalane Liquid Phase
Temp 45': Isothermal
Biphenyl Fraction - Silicone GumRubber Column
ERM$L TO 300.C AT 6OVYIN , ISOTHERMAL X)O*
' 185.C ' 300 4YIN
45 SEC
-.
Figure 4
I

Biphenyl Fraction
448
The low molecular weight compounds, soluble in isooctane and hot methanol,
were primarily biphenyl and 0-, m-, and p-terphenyl. These products were identified by
gas-liquid chromatography using a silicone gum rubber column (Figure 4 ) and a Carbowax
20M column with appropriate standards. The peaks were trapped and analyzed by infrared
and NMR spectroscopy for confirmation. Biphenyl was present in sufficient quantity to be
readily detected in the initial infrared spectrum and was isolated by suhimation. \
Quantitative data, obtained using the silicone column with appropriate calibration curves
are presented in Table 3. With the possible exception of biphenyl, the yields are very
low considering the overallconversion noted. The 0-, m- and p-terphenyl ratio (1/0.4/1)
indicates a preference for the para position beyond that expected for random attack.
-__ - -.- . -~ - - -_
If hydrogenation of the polyphenyls by hydrogen radicals generated "in situ" I
is to be considered a primary reaction mechanism, one would expect to see components
corresponding tothe possible hydrogenated species of biphenyl and the terphenyls . This
does not appear to be the case, however, as no significant peak can be ascribed to a
hydrogenated o-terphenyl compound. The small peaks before biphenyl and after o-terphenylt
in the chromatogram (Figure 4 ) actually include four or more, components present in \I
small proportions which have not been completely identified. The first peak, (eluted
before biphenyl) amounting to less than 0.1 per cent of the total yield, was initially
ascribed to a mixture of the possible hydrogenation products of biphenyl. On the Carbowax'
'>
column at lower temperatures this peak is eluted between phenylcyclohexane and I-phenylcyclohexene.
Initial infrared examination of this peak after trapping indicated an intriguing
similarity with the peak (s) noted after o-terphenyl and with the polymer fractions. The \
possibility that this component was a low molecular weight precursor to polymer formation
prompted further study. Infrared indicates a phenyl substituted aliphatic compound containing
some olefinic unsaturation, and the spectrum is not consistent with the anticipated
phenyl cyclohexenes . Careful examination of the spectrum indicates that the following
phenyl substituted compounds are not present on the basis of the indicated missing
absorptions - phenyl cyclohexane ( 1010, 1000, 888, 865 cm-' ), 1-ppylcyclohexene
(922, 805 cm-l), and 3-phenylcyclohexene (855, 788, 675 cm-l). The spectrum
is consistent with a non-conjuga'ted benzylcyclopentene system with absor tions at
1070, 1030, and 960 cm-' and the expected aromatic substitution bands.' Two
additional absorptions are noted, one at 850 cm-l which may be attributed to vinyl
protons and a phenyl substitution band at 730 cm-l. This likely results from biphenyl
contamination, but it is interesting to note that benzylidenecyclopentane has an absorption
at 732 cm-' with the remainder of the spectrum being similar to the benzylcyclopentenes.
NMR data provides strong evidence for a rigid ring system (non-olefinic
protons poorly resolved 1-3 ppm ), non-conjugated olefinic protons at 5.7 ppm and phenyl
protons at 7.1 ppm. The olefinic protons are complex, giving a closely-spaced doublet
superimposed on a broad absorption, suggesting the presence of both ring and exocyclic
unsaturation. Additional sharp proton resonances are noted at 1.2 ppm (methyl) and a
doublet at 2.7 ppm which is attributed to benzylic protons. The lack of olefinic protons
downfield from 5.7 ppm would rule out the benzylidene compounds and likely any sim-
ilar structures wherein -the unsaturation is conjugated with the phenyl moiety. The
olefinic protons in cyclopentadiene are found at 6.42 ppm and were not obvious in this
fraction. However, absorptions due to a minor component could have escaped detection
because of limited sample size. The difficulty encountered in isolating the components
of this peak precludes definite structural assignments, but the reaction of phenyl
radicals with fulvene followed by H radical termination and hydrogenation to give
~
4
I
'
,%
'i

j
,
i 449
i
1'
phenyl or benzyl substituted cyclopentenes is clearly indicated. The following, reaction
scheme is proposed to account for the experimental observations:
The inherent complexity of the product mix is illustrated by IV above which
could be 1-, 3-and 4:benzylcyclopentene. Preliminary mass spectral data for this sample,
as isolated from the chromatographic separation, indicates m/e values of 156 and 158,
I corresponding to the cyclopentadienes I, I1 and the cyclopentenes IV, V. Ring substituted
compounds such as I11 would also have a mass of 156. Mass 91, corresponding to the
benzyl radical is prominent as is mass 77, the phenyl radical, although this would be
!,'expected to arise from.the biphenyl (154) contamination. Other features of the spectrum
are being evaluated. The low yield is reasonable in view of the limited probability for
radical termination and hydrogenation compared with the high reactivity of these olefins
(or the diene precursors)'under the reaction conditions. The glow discharge work of
Schzler indicates the absence of any products with vapor pressure similar to biphenyl
and the other related literature generally does not @ve details beyond biphenyl analysis
and gross polymer properties .'
The peak after o-terphenyl (Figure 4) actually consists of. 3-4 components
when analyzed by varying the chromatographic conditions. This peak has the, same retention
time as diphenyl fulvene and is in the same general range as noted for a commercially
available hydrogenated terphenyl, Monsanto HB-40. This complex peak was trapped (with
trace o-terphenyl contamination) as a yellow liquid which gave an infrared spectrum
identical-to the polymer fractions and very similar to the peak.before biphenyl. Again,
phenylknzylcyclopentenes are indicated and the spectrum bears little resemblance to
the hydrogenated terphenyl spectrum. The NMR spectrum in carbon tetrachloride showed
aromatic protons at 7.1 ppm,, a broadened olefinic proton resonance at 5.7 ppm and non-
olefinic protons from 0.8-3.5 ppm with poor resolution. These extremely complex
olefinic and non-olefinic resonances are typical of protons in a fixed ring system such
as cyclopentene. The i,ntegration for these resonances allows approximation of an
average structure containing a cyclopentene ring system substituted with one phenyl and
one benzyl group. The reactions appear analogous to those responsible for the components
of the peak before biphenyl with termination being with a phenyl radical rather than the
hydrogen radical. The mass spectrum for this material as trapped from the chromatographic
analysis shows prominent peaks at mass 232 and 234 corresponding to VI and, VU below.
The benzyl and phenyl radical peaks are also evident at mass values of 91 and 77
respectively. Evaluation of the find structure details is continuing. The following is
believed to be representative of the many reactions which are possible under the
experimental conditions, to give a variety of closely related compounds:

LUd
The actual structural definition must await isolation of sufficient quantities of the com-
ponents for NMR studies under various conditions. The possibility of a hydrogenated
p-terphenyl was carefully considered, but the only structure even remotely consistent
with the infrared and NMR data would require that the center ring of p-terphenyl be
non-aromatic with m'ono-substituted phenyl groups attached. Such a material could be
formed by preferential hydrogenation of the center ring of p-terphenyl (statistically
unlikely ) or by the reaction of phenyl radicals and 1,3 -cyclohexadiene as follows:
This mechanism requires the formation of hydrogenated ortho terphenyl derivatives such
as VIII, which were not noted by gas-liquid chromatography or infrared.
TABLE 3
Biphenyl Fraction (a)
X
Yield %
Exposed Benzene
Biphenyl 0.30
o - Terphenyl 0.05
m - Terphenyl 0.02
p-Terphenyl
Benzyl and phenyl cyclopentenes (C12 1
0.05
0.05
Phenylbenzylcyclopentenes ( Cle) 0.25
TOTAL 0.72
VI1
(3)
Yield
% Reaction Products
3.5
0.6
0.2
0.6
0.6
-%
(a) Quantitative data were obtained by GLC, using a silicone gum rubber column with
appropriate standard calibration curves based on peak height.

Polymeric Products
The material which adheres to the glass dielectric surface in the reactor is
a high melting solid ( 3 320°), insoluble in benzene and all common solvents. The infrared
spectrum and the carbon-hydrogen ratio are essentially the same as noted for the benzene
soluble material. The benzene soluble polymer was fractionated into three molecular
weight ranges based on solubility in isooctane. The polymers are all yellow with the in-
tensity increasing as the molecular weight decreases. The ultraviolet spectrum for the
low molecular weight polymer shows a gradual tailing into the visible region. The
physical property data for the polymeric fractions are summarized in Table 4.
Infrared data indicate that the polymeric fractions are structurally similar
to the low molecular weight products identified as benzyl and phenyl substituted cyclopentenes.
The infrared evidence already presented for the low molecular weight products is applicable
to the polymeric products and need not be repeated. The data are consistent with an average
repeating unit containing the cyclopentene ring structure substituted with phenyl or benzyl
groups. NMR data for the polymers were obtained in carbon tetrachloride at the cell holder
temperature (40"). The spectrum obtained for the 300 molecular weight polymer fraction
is presented in Figure 5. The extreme broadening of the proton resonances is associated
with the complex, long range roton coupling in a risd system and the motional averaging
commonly noted in polymers .'2 Scanning the same sample at 90°-in tetrachloroethylene did
not significantly improve the resolution. The NMR spectrum of the- polymer in .pyridine
(Figure 5 ) gives some improvement in the high field proton resolution, indicating a doublet
at 2.6 ppm (benzylic protons ) and a complex methyl "proton resonance. The spectra are
similar to those obtained for the low molecular weight precursors containing unresolved ,
ring protons. The NMR spectra generally eliminate polymer formation by:way of phenyl
and hexatrienyl radicals as suggested for the radiolysis of benzene.23 The aliphatic
proton portion of the spectrum is very similar -to that reported for cyclopentadiene polymers
by Davies and Was~ermann.~~ The cyclopentadiene polymers had a molecular weight
range of 1200-2300, ah'max.at 320-360 mp and non-olefinic proton to olefinic proton ratio
of approximately 3/1.
The data are consistent with polymer formanon by way of phenyl radical
(or excited benzene) reaction with the fulvene produced to give phenyl or benzyl substituted
cyclopentadienes which then polymerize to gve a polycyclopentene chain with pendant
phenyl and/or benzyl groups. The average non-olefinic to olefinic proton ratio of 2.9
indicates that the many possible structures similar to XI (5/2) predominate over the
alternate type structures, XI1 (7/0), 24 assuming our analogy to cyclopentadiene type
polymers is valid.
Q
&,
XI
Many similar structures must be considered, including those derived from phenyl attack
on the ring with polymerization through the exocyclic vinyl group of fulvene.
(4)

I
I
I
I
I
In
m
9
In
2
,
I:
r-
In
m
\o
al
"0.
In
N
In
ln
Q\
\o
In
In
PI
hi
9
.-.I
-.
0
r-
In
cv
In
In
m
9
m
9
H
0,
9
c\
in
N
in
In
m
9
In
z!
m
P-
9
m
N
r-
m
i "
c
.d
\
\ I

453

’.
454
Mechanism I
The available evidence is thus not consistent with the generally proposed
i-iiechaii;si;i 0: i;~:j.~~i f~iii~atisii based ~ ithe i iaiidoiii bui:d-i;p d 2 PO:? (pheiijjl~iiej chain
1
accompanied by hydrogenation, or the interaction with the hexatrienyl diradi~al.~’ In the
radiofrequency discharge of benzene, evidence was presented which clearly indicated that
the polymer contained consecutive para linkages suggesting poly (p-phepylenes ) as the
predominant structure.12 Schcler observed that the polymer (C H 1.03, M.W. 503) was
a phenyl substituted aliphatic chain based on infrared evidence.16 Patrick and Burton have ‘
demonstrated that hydrogen atoms are not involved to any significant extent in polymer
formation when liquid benzene is irradiated with a 1.5 m.e.v. source. 24
\
In the corona discharge many types of energy transfer are occurring, varying
from photolysis (visible corona) to relatively high energy electrons responsible for frag- ‘i
mentation. The low yield of biphenyl in this single pass reactor suggests that phenyl radical
’
production through loss of a hydrogen radical is not the primary reaction route leading to
polymer formation. Considerable energy should be available to excite benzene to relatively
high vibrational levels which would be somewhat below that energy required to actually -\
separate. the H radical. At any given time the number of these excited benzene molecules
should greatly exceed the phenyl radical population. Fulvene is produced from benzene by
ultra-violet energy (200 mp, about 112 kcals, or 4.9 electron volts).25 Thus, the formation
of this benzene isomer with a resonance energy (11-12 kcals/mole) intermediate between
benzene and 1,3 cyclohexadiene may be energetically favorable in the corona environment. \\ 1
Fulvene has been shown to polymerize rapidly under similar conditions with no reversion
to benzene.25 The uniformity of the phenyl (benzyl) -cyclopentene ratio throughout all
polymer fractions makes it difficult to accept a mechanism based solely on random attack 1
by phenyl radicals on the growing fulvene polymer. The initial synthesis of a monomeric
unit comprised of the benzylcyclopentadiene system with subsequent diene type polymerization
is consistent with all our observations. The phenyl radical produced in the discharge
should collide with the nearest benzene molecule which will be excited to a \
relatively high vibrational energy level. The following reaction sequences are suggested
for the phenyl radical:
U
Re*=+
Reaction (5) is the typical termination reaction by radical coupling leading to biphenyl.
The intermediate radical in (6) would be expected to lose hydrogen mo e readily than
it would add to another benzene molecule to give polymer formation.2d The possibility
of ring contraction in the excited intermediate, accompanied by an intramolecular shift
of a hydrogen atom is suggested in reaction (7). The benzylcyclopentadiene XV so
obtained would be a very active monomer leading to the polymeric products observed.
Although our data support polymer formation by the mechanism indicated, we cannot
rule out participation by the cyclohexadienes and acetylene, both of these being noted
a - e n e - CkEketewr-eentper-at-.
XIV

1'
b
Scheme I
fulvene
455
FIGURE 6 PROPOSED REACTION MECHANISMS
I
iHO
1,4 1.3
biphenyl
A\ biphenyl '. 0.- terphenyls
cyclohoxodienes cycloherene cyclohexane
@z, /I P ) 7 3 C H 3
. . .. . . .. . benzyl cyclopentenes
/ \ 05CH3-
phenyl, methyl
substituted
cyclapentenis
phenyl, benzylrubstlhlted
u cyclopentadiene type polymers
/

456
, , .
The relatively low yield and the complexity of the reaction products does not
allow an unequivocal designation of reaction mechanism or of the mode of energy transfer in
the corona environment. However, where possible. we have proposed reaction mechanisms
which are consistent with our results to date. These mechanisms are summarized in Figure
6. Studies are underway to obtain deuterium exchange data in the corona reactor, and mass
spectral data are being evaluated. Investlgation of the utility of corona chemistry will be
' I
extended to other volatile compounds.
EXPERIMENTAL
Materials: Fisher thiophene-free benzene contained trace quantities of
methyl cyclopentane, cyclohexane and toluene. Chromatogram (squalane ) of original
material was used as reference for irradiated benzene evaluation under the same GLC
conditions.
Reference materials for chromatographic identification were generally used
as obtained from commercial sources, Most of these materials were of sufficient purity to
identify the major peak. However, mono-phenyl fulvene, obtained from Aldrich, gave no
peak on the silicone column and infrared indicated extensive hydroxyl and carbonyl ab-
sorptions. 1,3,5-hexatriene from Aldrich gave two major and two minor peaks on the
squalane column. Infrared27 indicated largely the transisomer with only a small cis ab- )
sorption at 818 ern-'. 1
Fulvene was prepared by the method of Meuche. l7 Infrared (1662, 925,
890, 765 cm-l) and ultraviolet (A1 = 242 mu, 4 2 = 360 mu) were in good agreement
literature values: '5 '
Reaction Conditions
The equipment used in these experiments was presented in Figures 1 and
2. Helium was bubbled through the benzene reservoir (55") at 100 ml/min. to carry 4.2
~(0.054 moles) benzene through the corona reactor per hour. The reactor temperature
was 45": 2". A total of 101 gm. (1.3 moles) benzene was passed through the 15 KV, 60
cycle, corona discharge in 23.7 hours. A small helium purge was maintained throughout
all sampling operations and necessary downtime to preclude oxidation.
Product Isolation and Analysis !
At the conclusion of the run, the reactor was partially filled with benzene
and refluxed on a steam bath for 2 hours. ?he berizene.soluble material from the column
was combined with the yellow brown solid which collected in the lower receiver. The
insoluble material was swelled by bekene and generally loosened from the glass di-
electric surface. This insoluble material was filtered off, washed several times with
chloroform and dried overnight at 70" in a vacuum oven. The reactor was cleaned
with.a 5 per cent hydrofluoric acid solution to remove final traces of polymer after
each run. Benzene insoluble polymer - 0.7 gm. yellow brown color, % C 86.95, %H
7.17, C/H 1.01.
The benzene containing all solid products from the reactor and pot
residue was reduced to 30 ml. volume by vacuum distillation at 50". Considerable
foaming was encountered, requiring careful control of the distillation. The 30 ml. Of
I
'
I
\
\
\
\
,
\
\
i

- . .. . - . . . -
457
brown benzene solution was added to 300 ml. isooctane at room temperature. A yellow solid
precipitated immediately and was collected by filtration. After several washings with hot
methanol, the material was dried for 8 hours at 70° in a vacuum oven. Benzene soluble
polymer - Fraction 1, 2.3 gm., M. W. 4360, %C 88.32, %H 7.18, C/H 1.02.
The isooctane-methanol solution was reduced to 100 ml. volume (all
methanol removed) by vacuum distillation at 60". The solution was a bkilliant yellow
color at this point. On cooling to room temperature, a small proportion of a deep yellow
polymeric solid precipitated and was collected by filtratlon. After washing with hot
methanol, this yellow material was dried at 70Ofor 8 hours in a vacuum oven. Benzene
soluble fraction 2, 0.2 gm., M.W. 1555, %C 87.26, %H 7.09, C/H 1.03.
The isooctane-methanol soluble portion was reduced to 20 gm. by vacuum
distillation and sufficient benzene added to insure solubility of ell components. The per
cent solids was determined for this sample, being careful to.limit sublimation of biphenyl
(about 2 hours at 70° in a vacuum oven). This sample contains the low molecular weight
products and a yellow resinous polymer. Qualitative analysis for the low molecular weight
components was accomplished by gas-liquid chromatography using the following liquid
phases: silicone gum rubber, Carbowax 20 M, and Reoplex 400. Quantitative data were
obtained on the silicone column (Figure 4) using calibration curves (peak height) prepare-d
from appropriate standards. The concentration of the small peak eluted before
biphenyl was estimated using the biphenyl calibration curve and the peak after'o-terphenyl
was estimated using a commercial hydrogenated terphenyl mixture (Monsanto
HB-40) as a reference. The F & M 500 gas-liquid chromatograph equipped with a
standard thermoconductivity detector was used for these analyses. These low molecular
weight products (biphenyl, 0-, m-, p-terphenyl.and phenylcycloalkenes, M. W. up
to -250) amount to 0.72 gms. (See Table 3). Peak confirmation was provided by infrared
and NMR analysis after fraction collection from the chromatographic separation.
The concentration of the low molecular weight resinous polymer was
determined by subtracting the total biphenyl fraction weight from the total isooctane-
methanol soluble material. A sample of this material was placed in the vacuum oven
at 100" unnl the GLC trace indicated neglipble biphenyl content and the molecular
weight was determined. Benzene soluble fraction 3, 3.5 grn., M. W. 305, %C 87.23,
%H 7.09, C/H 1.03. Prolonged heating at 150" in air gwes a hard, brittle yellow film.
Infrared analysis indicates oxidation is involved in the drpng process.
The benzene fraction was initially examined by GLC using 100 ft.
support coated open tubular columns containing Carbowax 1540 poly (ethylene glycol )
and squalane (Figure 3) as the liquid phase. Quantitative data were obtained at 45O
on the squalane column with a helium flow of.3.0 ml/min. The Perkin Elmer Model
2.80 gas-liquid chromatograph, equipped with a stream splitting device and hydrogen
flame detector was used for these analyses. Peak identification was by comparison
with commercial standards and the synthesized fulvene. Additionally, the beniene
fraction was refluxed in pressure bottles with maleic anhydride, azobisieobutyronitrile,
and ammoniacal cuprous chloride solution prior to chromatographic analysis to note
peak changes due to adduct formation, polymerization or presence of acetylenic hydro-
gen. Acetylene content was estimated from weight loss of the -7O"and -195" traps.
The vapor space above those cold traps and the gas stream directly below the corona
were analyzed by infrared using an 8 cm. gas cell. Although the study was not
exhaustive, several samplings were made and only acetylene was detected.
'

458
h/lethylacetylene, allene and butadiene were not present in sufficient quantity to be detected.
The yellow ,benzene from several runs ( 1000 mlt ) was fractionated and
900 mi. distillate obtained from 79-80". The distillate was refluxed with maleic anhydride
until colorless and the benzene removed at reduced pressure. Dilute sodium hydroxide
solution was added and the resulting solution was extracted several times with chloroform.
After acidification with dilute hydrochloric acid, the solution was extrayted with ether.
The et1ie.r was removed after drying to give a white solid which, on recrystallization from
chloroform and pet ether, gave the adduct 7-methylene-5-norbornene-2, 3-dicarboxylic
acid \VI ich melted at 146°-1500. The infrared showed maxima at 1553, 875 and 710 cm-',
(lit ("' 1555, 874, 712 cm-'). Fulvene was prepared in low yield (less than one per cent)
from cyclopentadiene and formaldehyde with sodium ethoxide catalyst. l7 The adduct was
prepared by adding maleic anhydride to the crude fulvene reaction product in Freon I13
(b.p. 47.6O) and refluxing until colorless. After removal of the solvent, hydrolysis and
recrystallization yielded the adduct, m. p. 147-150°, which gave properties identical with
the fulvene adduct obtained from the corona discharge. Bryce-Smith obtained a melting
range of 105-110° for this adduct which was likely a mixture of the ex0 and endo isomers.25
Stille prepared the fulvene-maleic anhydride adducr with a melting point of 149-150" which
may be a single isomer.12
Through the courtesy of the Perkin Elmer Company at Norwalk, Con-
necticut, the discharged benzene was analyzed by combination of gas-liquid chromatography
and mass spectroscopy. The peak attributed to fulyene gave a parent mass of 78 and
fragments at m/e 77, 52, 51, 50 and 39 in good agreement with the literature. I' Con-
firmation of other assignments was obtained by this technique and the one unidentified
component on the squalane column gave a parent mass of 78.
Molecular weight and carbon-hydrogen determinanons were made by
Schwarzkopf Microanalytical Laboratories, New York, New York. Molecular weight
was determined in benzene by vapor pressure osmometry.
LITERAT-URE CITED
1 Present address: Tarrytown Technical Center, Union Carbide, Tarrytown, New York.
2 G. Glockler and S. C. Lind, "The Electrochemistry of Gases and Other Delectrics,"
John Wiley and Sons, New York (1939).
3 Ozone Chemistry and Technology, Advances in Chemistry Series No. 21, American
Chemical Society (1959).
4 J. A. Coffman and W. A. Browne, Scientific American, 89, June (1965).
5 M. Bertholot, Compt. rend., 82, 1357 (1876).
6 S. M. Losanitsch, Bull, SOC. stun. Bucharest, 233 (1914).
7 A. P. Davis, J. Phys. Chem., 35, 3330 (1931).
8 V. Karapetoff, W. H. Trebler, E. G. Linder, and A. P. Davis, Research Reports,
Detroit EdisOn Company, Cornel1 Univ. (1928-32).
9 G. P. Brown, R. E. Rippere, Am. Chem. SOC., Dv. Petrol. Chem., Preprints 2,
NO. 3, 149-54 (1957).
10 H. SchUler, K. Prchal, and E. Kloppenburg, Z. Naturforsch., 15a, 308 (1960).
11 A. Streitwieser, Jr., and H. R. Ward, J. Am. Chem. SOC., 84, 1065 (1962), 85,
539 (1963).
12 J. K. Stille, R. L. Sung, and J. Vander Kooi, J. Org. Chem., 30, 3116 (1965).
13 U. S. I. Chemicals, Polyolefin News, Feb. (1964).

459
14 G. J. Jam, J. Chem. Phys., 22, 751 (1954).
15 J. Thiec and J. Wiemann, Bull. SOC. chim., France, 177 (1956).
16 J. M. Blair and D. Bryce-Smith, Proc. Chem. SOC., 1287 (1957).
17 D. Meuche, N. Neuenschwander, H. Schaltegger, and H. V. Schlunegger, Helv.
Chim. Acta, 47, 1211 (1964).
18 G. E. Gibson, N. Blake, and M. Kalm, J. Chem. Phys., 21, 1000 (1953).
19 G. F. Woods and L. H. Schwartzman, J. Am. Chem. SOC., 70, 3384 (1948).
20 E. E. Van Tamelen and S. P. Pappas, J. Am. Chem. SOC., 85, 3297 (1963).
21 E. L. Eliel, J. W. McCoy, and C. C. Prlce, J. Org. Chem., 22, 1533 (1958).
22 J. G. Powles, Polymer, 1, 219 (1960).
23 W. N. Patrick and M. Burton, J. Am. Chem. SOC., 76, 2626 (1954).
24 A. G. Davies and A. Wassermann, J. Polymer Sci., 4, 1887 (1966).
25 H. J. F. Angus, J. M. Blair, and D. Bryce-Smlth, J. Chem. SOC., 2003 (1960).
26 G. A. Mortimer and L. C. Arnold, J. Am. Chem. SOC., 84, 4986 (1962).
27 J. C. H. Hwa, P. L. de Bonneville and H. J. Sims, J. Am. Chem. Soc., 82, 2538 (1960).

Introduction
VAPOR PHASE DECOMPOSITION OF
AROhIATIC 'HYDROCARBONS BY ELECTRIC DISCHARGE
hlasayuki Kawahata
Research and Development Center
General Electric Company
Schenectady, New York
Decomposition of organic compounds in presence of electric dis-
charge generally involves fragmentation and polymerization induced by
inelastic collision with high energy electrons. When hydrogen is
added, active hydrogen atoms produced by electric discharge also par-
ticipate in the decomposition reactions. The reaction mechanisms are
complex involving excited molecules, free radicals, and ions.
In this laboratory hydrocracking of coal, coal volatiles and related
materials by electric discharge in hydrogen was studied. It is
postulated by Given1 that coal (vitrinite) molecules contain aromatic
and hydroaromatic structures and probably fused aromatic ring nuclei
linked together by methylene or ethylene groups forming hydroaromatic
rings. Many of the replaceable hydrogens in the structure are substituted
by hydroxyl or carbonyl groups. Short alkyl groups and
alicyclic rings may also be attached as side chains. In pyrolysis
at 500 - 6OO0C., dissociation of hydroxyl groups and dehydrogenation
of naphthenic rings take place. These acts lead to formation of OH
t
and H radicals which in turn help to break the linkages between aromatic 1
nuclei, forming smaller, partly aromatic volatile fragments, and at
the same time leaving a more aromatic residual structure behind.
On the other hand, high pressure hydrogenation of coal gives partial'
or complete hydrogenation of the aromatic structure. When this is sub-
sequently cracked, the light products tend to be aliphatic rather than
aromatic hydrocarbons. It was thought that hydrocracking of coal by
electric discharge might be somewhere between pyrolysis and high pressure
hydrocracking. In the course of study it was thought important to test
this hypothesis by subjecting several aromatic or hydroaromatic compounds
to electric discharge in a hydrogen stream for better understanding of
the process. Organic compounds selected were m-cresol, d-methylnaphthalen
tetrahydronaphthalene, decahydronaphthalene, and 9, 10-dihydrophenanthrend
Experimental Apparatus and Procedures
The apparatus consisted of a vaporizer, a preheater, a discharge
reactor, and a product collecting system: The organic compound was fed
into the vaporizer at a certain rate and vaporized. Hydrogen was also
P. H. Given, presented Am. Chem. SOC., January, 1963 in
Cincinnati, Ohio
\
\
\
'6 1
1
?
1

461
fed into the vaporizer at a certain flow rate and mixed with the organic
vapor. The resultant mixture was then fed into the reactor through
the preheater. The liquid products condensed in a water cooler were
collected in a receiver and the gaseous products were collected in a
liquid nitrogen trap. These products were analyzed by mass spectro-
scope and vapor phase chromatograph employing a 6 ft. column of 10%
silicone rubber, SE-30 and 60-80 Chromosorb P.
I
The discharge reactor was fabricated with quartz and was of a
concentric tube design similar to an ozonizer. The inside of the inner
barrier (40 mm OD and 38 mm ID) and the outside of the outer barrier
(48 mm OD and 46 mm ID) were coated with conductive, transparent, tin
oxide. The former was connected to the high voltage terminal and the
latter to ground. In this arrangement the electric discharge was
sustained in an annular space, 46 mm OD, 40 mm ID, and 200 mm long.
The inside barrier tube was filled with stainless steel wool and a
thermometer was placed in the center. This thermometer and three therm-
istors attached to the outside electrode were used to determine the
reactor temperature. The reactor was insulated with glass wool, and the
reactor temperature was maintained at approximately 30OoC. for all the
runs.
The electric discharge power was supplied by Peeding the output
of a 10,000 Hertz, 30 kilowatt inductor-alternator to the primary of
a 50 kilovolt transformer and, in turn, to a tuned circuit, to the
reactor and to the high voltage instrumentation. The electric discharge
power was determined by measuring the area of parallelogram on the
oscilloscope2.
When making the run, hydrogen was fed into the reactor at a definite
flow rate and the electric discharge was applied to heat the reactor.
When the reactor temperature was stabilized at about 3OO0C., the feed
of the organic compound was started. The discharge power sustained for
the reaction was in a range from 140 to 170 watts.
Experimental Results and Discussions
For cresol-hydrogen mistures, two runs were made using the empty
reactor and three runs were made by filling the reactor space with
porous or activated aluminum oxide grains. For all the runs, the reactor
pressure maintained at 300 mm Hg. Experimental results, including
product distribution, are listed in Table 1. The principal products
were phenol, toluene, benzene, aliphatic hydrocarbons, carbon dioxide
and water. Among the aliphatic hydrocarbons, acetylene was present in
the largest amount and about 8% was unsaturates except for Run 5. In
this run the concentration of unsaturates was 55 %. This is probably
due to a higher hydrogen concentration in the feed causing somewhat
T. C. Manley, Trans. Am. Electro Chem. SOC. 84 83, 1943

462
more efficient hydrogenation. The bond energies of C6H5-CH3 and
C H5-OH are 90 and 73 Kcal/mol. respectively. Despite this, it was
otserved that, in the products‘. phenol was in higher concentration
than toluene.
Use of aluminum oxide packing in the discharge space was intended ,
to investigate the possibility of increasing the energy yeild. Narrowing
the gaseous discharge gap with dielectric packings may cpuse the follow- \
ing two effects on the discharge: (1) increase of discharge current
for the same discharge power dissipated, and (2) increase of gaseous
space breakdown field strength, if the gap decreases beyond a certain
l i m i t . The exact nature of the electric discharge employed in this study
is still debatable.3,4
However, it can be reasonably assumed that the )
primary reaction rate of the organic vapor with either high energy
electrons or active hydrogen atoms may be dependent on discharge current
density and field strength, if the system pressure and partial pressure
of the reactant are constant. In electric discharge product of ozone.
an increase in ozone concentration after filling the discharge gap with
various dielectric packings is also reported by Morinaga and Su2uki.S
In this study for approximately the same concentration of cresol vapor,
the use of aluminum oxide grains in the discharge space appeared to in-
crease the energy yield somewhat, but not conclusively.
In all the runs, the formation of brown solid films was observed
on the reactor wall or on the surface of the grains. These solid films
were not analyzed but they were insoluble in methylethyleketone or 1
I
toluene. It is presumed that they are indicating possibly highly cross-
linked polymerized products derived from the cresol. The energy yield
for film formation was estimated to be in a range from 30 to 50 g/KWH;
considerably higher than that for the fragmentation products.
Experimental results for the polycyclic compounds are summarized
in Table 2. Methylnaphthalene vapor in hydrogen was tested under pressures
of 760 and 74 mm Hg. The principal lighter products were aliphatic
hydrocarbons, benzene and toluene. At the higher pressure the concentration
of the lighter aliphatic hydrocarbons was in the order C2)c3>C4>C5. ~
A t the lower pressure, however, this order was reversed. In both cases ,\
about 32 - 35% were unsaturates. The energy yield was approximately
doubled by lowering the pressure.
For tetrahydronaphthalene, three experiments were made under diff
erent pressures. Among the 1 ighter aliphatic hydrocarbons produced,
the C2 fraction was present in the largest amount, in which ethylene
was in highest concentration followed by acetylene and ethane. The
)
total percentage of unsaturates increased as the pressure decreased.
This increase was essentially due to the increase in C2 and C un-
7
saturates. As observed for methylnaphthalene, the energy yeiqd at .:
70 mm Hg was twice as high as that at 760 mm Hg. A considerably higher
R. W. Lunt, Advanced Chem. Series, “Ozone Chemistry and Technology”,
Am. Chem. SOC., p. 286 (1959) 1
M. Suzuki and Y. Naito, Proc. Japan Academy, 2 469 (1959) \
K. Morinaga and M. Suzuki, Bull. Chem. SOC., Japan 35 429 (1962)
\
‘1 I
\
j
‘I
4
1
I
\

i
453
energy yield than for the other two runs was observed at the intermediate
pressure, 300 mm Hg.
1 760 mm Hg using 6.7% concentration in hydrogen. Another run was made
When decahydronaphthalene was tested, one experiment was made under
under 300 mm Hg using 67% concentration. In the latter run, a higher
/energy yield was obtained. The product distribution was richer in the
c2 and C3 fraction and also unsaturate concentration wab higher.
'
One experiment was made to test dihydrophenanthrene in hydrogen.
The gaseous products were essentially in the C2 fraction; about 50%
being unsaturated. Vapor phase chromatography of the liquid product
showed two peaks. They were not identified but by the combined results
of the VPC and mass spectroscope one of the peaks was tentatively
I identified as butflbenzene. Biphenyl, which is a likely decomposition
' product, was not found. For the polycyclic aromatic and hydroaromatic
compounds tested in this study, the energy yield from electric discharge
hydrocarcking for production of the lighter hydrocarbons was in the following
order:
/
I
I
I The number of experiments are not sufficient to permit drawing a
concrete relationship between energy yield and molecular structure.
As observed in cresol runs, solid films were formed on the reactor W a l l ,
but they were not analyzed.
The resonance energies of benzene, naphthalene, and phenenthrene are
39, 75, and 110 Kcal, respectively. It is reasonable to assume that
' the condensed aromatic ring structure absorbs large amounts of energy
and requires high energy for cracking. The radiation effect on various
polycyclic aromatic compounds were studied by Weiss et a16. They dis-
,\ cussed correlations between the radiation stability and various strut-
$ 1 tural factors. These include resonance energy, electron affinity,
I and ionization constant. Since there is a close similarity between
radiation and electric discharge in principle, further information along
this line would be helpful for a better understanding of the electric
discharge hydrocracking process.
b
Elucidation of the reaction scheme in detail is beyond the scope
/of the this study. However, it was indicated that in electric discharge
hydrocracking of aromatic or hydroaromatic hydrocarbons, dissociation
of the side chains and rupture of the rings are followed by secondary
freactions involving the decomposed species; this leads to the formation
of aromatic or aliphatic lighter compounds. Energy requirement to form
these lighter compounds appear to be too high for practical applica-
[ tions. The formation of the solid polymerized products, which takes
\
1
I
\
v ' ~
$
J. Weiss, C. H. Collins, J. Sucker, and N. Carciello, Ind. Eng. Chem,
Prod. Res. and Development, 3, 73 (1964)

464
place in parallel with fragmentation, as seen for cresol runs, requires
considerably less energy. Study on the formation of polymer films
starting with various monomers and using the present discharge system
presents an extremely interesting problem which is currently under
investigation.
This is a part of the work supported by the Office of Coal Research,
The author is thankful fpr their generous\
U. S. Department of Interior.
support .
1
\
\

Ldd
ffl
cc
3 .I+
X
z
\rl
M O
C
- 0
fflc
e a
0
k
&.2 u”
vc,
ldv
c E\
ld o z
M k
k 4 u
0 u
u)
kM
oc
c, .I4
O X
ldv
alld
m a
0
2
a
X
W
In
W
0
IC
hl
m
0
rl
m
hl
In
0
0
E
In
In
(0
hl
In
In
In
IC
rl
m
In
r(
rl
hl
a3
0
0
a0
m
m
a
al
ld c,d
m c ldc
I n c
U a b
m c o
w - M
46 5
rl
0
,ul
al
k
V
A
y-c
0
u
d
k
c,
V
al
rl
W

466
P
rl
0,
k
7
rnI
rn
al
k
a
I
hl
m
b
m
l-l
I-I
hl
w
hl
m
m
hl
(D
(D
0
0
(0
t-
In
w
v)
m
N
0
*
Q,
N
*
N
0
m
N
r(
*
b
m
ro
w
b
hl
l-l
m
hl
rl
b
b
In
In
rl
0
(D
b
In
In
m
hl
d
m
(D
d
m
IC
rl
m
m
0
0
m
m
t-
0
Q,
In
In
hl
m
hl
*
b
I
I
0
m
0
b
w
b
Q,
*
rl
CJ
b
0
d
0
@a
a0
m
*
Q,
m
0
W
b
b
W
0 4
4 4
W
b
b
hl
b
b
hl
(D
ua
*
9
In
0
0
m
b
(0
(0
*
*
rl
t-
rl
@a
m
t-
m
N
m
Q,
(D
0
0
W
b
In
*
al
k
c
V
.d
c
3
m
c
0
.A
c,
V
cd
k
w
k
0,
.A
>
cd
0)
c
al
c
c,
m
cd
3
h
$
(D
c
7
k
I/)
.A
c
c,
k
0
w
c
0
.A
c,
1
P
.A
k
c,
rn
.A
5
c,
V
7
P
0.
kU
aa
>
al
Ch
W P
+la
oa,
.A
al'u
04
EIc,
dfi
dal
dU
alc'
co
bfi
v
nlc
*
P
fi
cd
V
.A
c,
d
Em
O V
kfi
41
0
v a
E
4 0
hU
u
A V
d .I4
Oc,
a d
E
rn wo
.E O h
0 (d
.A M O
c, Eh
u dQ
cd Y h
k OX
w d
k
Q, V
0
k
U
I
rl a
V h
X
fi
.A
m
.a,
c,
cd
k
1
c,
cd
m
fi
1
'u
0
a,
M
cd
c,
fi
0)
0
k
a,
a
*
*
I
a'
I
1

HYDROCARBONS AND CllRBON FROM A ROTATING ARC HEATER
C. Hirayama and D. A. Maniero
Westinghouse Electric Corporation, Pittsburgh, Pa.
INTRODUCTION
A brief description of the rotating arc heater utilized in this work,
and some preliminary hydrocarbon processing data, were described a year
ag0.l A more detailed description of the hbater, and some of the operating
characteristics have also been presented within the past year.2 The
arc heater consists essentially of water-cooled, toroidal electrodes, in
which the arc is rapidly rotated as a result of interaction of the arc
current with a magnetic field. The heater is a high current, low voltage
device with a rating of 3 megawatts into the arc. Both alternating and
direct current power operation are possible with this device.
In contrast,
most arc devices previously reported3 8pe of the long arc, high voltage
type which are operated on d.c. only.
We wish to report, herewith, some of the res.ilts obtained more recently
in the pyrolysis of methane at atmospheric and at slightly elevated pressures.
The feed gas was commercial grade methane of 96 per cent purity which
was injected into the heater at ambient temperature. The flar rate was controlled
by regulating the pressure drop across a sonic orifice.
The arc heater was operated only on a.c. parer at electrode separations
of 0.38, 0.75 and 1.0 inch. The arc power with methane ranged from 550 to
2200 kilowatts, with thermal efficiencies between 25 and 76 per cent, depending
on operating conditions. For operation at elevated chamber pressure,
the nozzle was choked with a graphite plate which had an orifice of 0.5 or
0.75 inch diameter in the center of the plate. The chamber pressure was as
high as 100 psig.
The heater was operated at two different field coil currents to determine
arc rotation velocity on the degree of reaction.
The products of the reaction were quenched and collected through a
water-cooled copper probe, 118 or 1/4 inch I.D., inserted about an inch
inside of the nozzle. The product from the probe was first passed through
'
a fiber filter (Purolater Co.) to separate the carbon, then collected at
appropriate intervals in ga8 sapling tubes on a manifold, a8 previously
described .1
The gas analyses were made mas8 spectrometrically. A typical composition
of a sam3le is shown in Table I; Table I1 shows the approldmate material
balance based on the &/C ratios of the feed and product gases. The only
variation with operating conditione is In the relative concentration of
the different species.

458
TABLE I. COMPOSI'IION OF GASEOUS PRODUCT, MOU PERCENT
~ ~ a c y 4 ~ ~ ~ c &
77.68 1.27 0.09 5.81 13.58 1.01 0.12 0.32 0.14
TABU 11. MATERIAL BALANCE OF PRODUCTS IN MOLE PERCENT
c o c q cy4 sc&='ac
73.06 1.19 0.08 5.46 12.77 0.95 0.11 0.30 0.13 15.00
Electron micrographs were obtained on a number of the carbon samples
by replicating an amyl acetate suspension on a carbon film. X-ray spectra
were obtained by packing the soot into a disc-sarnple holder and irradiating
with Cu K-a radiation. The sample packing and irradiation were kept as
nearly identical as possible for all of the samples. The d-spacing and line
intensity were compared against AUC graphite measured under identical conditione.
RESULTS AM, DISCUSSION
As shown in Table 11, the product consists of species which are expected
from the high temperature pyrolysis of methane. The primary products are H2,
C2H2, and carbon, with varying concentration of C&, depending on the degree
of reaction. The diacetylene and benzene axe fonned from the polymerization
of the acetylene; the methyl acetylene, although not intermediate in the
conversion of to C2H2, is known to be a minor product found in the pyrolysis
of diacetyleney The C,H, is an intermediate in the C'&+ C,H, reaction.
There is a slight error in the material balance since the hydrogen content
of the soot was not taken into account because of sampling difficulty. The
soot usually contains about one percent of hydrogen, and the presence of
aromatic constituents is evident from the odor.
Considering the difficulty of obtaining exact experimental parameters
during an experiment, quite good correlation was obtained between the degree
of reaction and arc enthalpy, whereas the correlation was poor with respect
to the net enthalpy increase of the gas. Figure 1 shows the relative degree
of methane pyrolysis as a function of the arc enthalpy. The curves show the
comparison between runs made at 2500 and 1500 amperes field coil current.
The former effects an arc rotation velocity approximately 66 percent greater
than that at the 1500 amps field coil current. There is a significantly
steeper slope at higher arc rotation velocity, this result probably arising
from the greater degree of mixing.
Note that the curyes saturate at high
enthalpies, where the system is at equilibrium, or near equilibrium condition.
A minimum just above the critical enthalpy (or initiation temperature).
The electron micrographs of the soot, quenched at the heater nozzle,
,500X magnification, Figure 2, shows spherical particles of less than
100 at 2l to approximetely 5000 A diameter. The nature of the background is not
knonw at present, but blow-ups of the photomicrographs to approximately
153,OoOX suggests that the cloud consists of extremely fine, smokey carbon
dust.

J
i
I
i
I
469
The d-spacings ob ained from the x-raa diffraction of the soot varies
between 3.45 and 3.49 B ,ocompared to 3.35 A for pure, crystalline graphite.
The lower value of 3.45 A was obtained on the carbon collected from the high
pressure runs, where the residence times were up to seven times longer than
the runs at atmospheric pressure. The degree of crystallinity of the soot,
based on the x-ray intensity, varied Prom two to ten percent of:AUC graphite.
The x-ray crystallinity, as suspected, is a function of the residence time.
It was of interest to plot the major product compositions on a semilog
plot, as a function of reciprocal arc enthalpy since the latter is proportional
to the temperature in the gas. Figure 3 shows that the slopes
for acetylene and hydrogen are approximately the same, whereas the slope for
carbon is much steeper. The difference in slopes for carbon, as compared to
CBH2 and H, clearly shows the different mechanisms for the formation of these
species.

REFERENCES
1. C. Hirayama and D. A. Maniero, "A New Rotating Arc Heater for Chem-
ical Processing," Preprints, Div. of Fuel Chem., Am. Chem. SOC.
Meeting, Pittsburgh, March 22-31, 1966, Vol. 10, No. 1, pp. 123-132.
2.
D. A. Maniero, P. F. Kienest, and C. Hirayama, Westinghodse Engineer
- 26, 66 (1966). Also, same authors, paper presented at Electrochem.
Sac. Eat. Meeting, Phila., Oct. 12, 1966.
3. See, for example, S. A. Miller, "Acetylene, Its Properties, Manufacture
and Uses." Vol. I, Academic Press, New York, 1965, pp. 391-405.
4.
K. C. Hou and H. B. Palmer, J. Phys. Chem. 69, 858 (1965).
tj
I

Arc Enthalpy ,-)
- 2500 Amps Field Coil Current
-- 1500 " 11 It I1
Fig. 1. Methane conversion as functions of arc enthalpy and arc rotation
Fig. 2. Electron micrographs of carbon at 25,500X

i
472
Reciprocal Arc Enthalpy ,->
Fig. 3. Product composition as a function of
reciprocal aro enthalpy

473
THE REARRANGEMENT OF METHANE AND METHYL CHLORIDE IN A MICROWAVE
DISCHARGE
I. INTRODUCTION'
J. P. Wightman
Virginia Polytechnic Institute, Blacksburg, Virginia
N. J. Johnston
NASA - Langley Research Center, Hampton, Virginia
In a previous paper(1) the rearrangement of a series of
hydrocarbons including methane in a microwave discharge was reported.
The work reported here is an investigation of the effect of the
introduction of a functional group in methane on the nature of both
the gaseous and solid products of rearrangement. Such measurements
should ultimately support a model of the rearrangement mechanism
in a microwave discharge. Methane and methyl chloride werc passed
separately in the absence of a diluent through a microwave dis-
charge.
11. EXPERIMENTAL
A. Materials
Methane (ultrahigh purity grade) was obtained from the
Matheson Co. and used without further purification. The following
analysis was supplied with the methane: C02 - (5 ppm; O2 5 ppm;
N - 19 ppm: C2H6 - 14 ppm: C3H8 - (5 ppm. Methyl chloride
(gigh purity grade) was obtained from the Matheson Co. and used
without further purification.
B. Apparatus and Procedure
The rearrangements were carried out in a high vacuum
flow system. The power source for the microwave discharge was a
Raytheon generator(Mode1 KV-1?4) and was operated at a power level
corresponding to about 40 r.f. watts at 2450 Mc. The generator
was connected to an air-cooled cavity(Raytheon - KV series) by a
coaxial cable. Two procedures were used depending upon whether
an analysis of the gaseous products was being made or whether solid
film was being deposited for subsequent analysis. In the former
case, the parent gas was passed through a variable leak valve
(Veeco - VL). The pressure in a ty ical experimental run was 0.15
torr at a mass flow rate of 1 x lo-' moles/min. and a linear flow
rate of 20 cm./sec. In the latter case, the parent gas vas passed
through a Teflon needle valve (Fischer-Porter) . The pressure in a
typical experimental run was O.5Oitorr. Normal deposition time
was 20 min.
The gaseous rearrangement products of methane and methyl
chloride were determined using a mass spectrometer(Associated
Electronics Industries - MS-10). The sampling volume was located
about 140 cm. downstream from the discharge. After establishment
of a constant flow rate, the parent gas stream was sampled. An
off-on valve upstream from the discharge allowed the system to

474
be pumped down without altering the setting of the variable leak.
Between analyses pumpdown resulted in faster attainment of steady
state concentrations of rearrangement products. Flow was again
started and the discharge initiated with a Tesla coil. After
establishment of a constant flow rate the discharged gas stream
was sampled.
The solid polyermic films were removed mechanically
from the walls of the Pyrex tubing. The infrared spectra of the
films of both gases were obtained using either a Perkin-hlmer
421 or 621 spectrophotometer. The electron spin resonance spectra
of the neat films from both gases were obtained on a Varian ESR 6
spectrometer. Elemental analyses of both films were made by
Gailbraith Labs.
111. RESULTS
’ A. Gaseous Products
1. Mass spectra
Selected peaks from the mass spectra of methane and
methyl chloride obtained with the discharge on and off are shewn
in Figures la and lb. The most significant features ip the methane
spectra(Figure la) were the incrTase in the m/e = 2 (H- ) peak and
the decrease in the m/e = 15 (CH3) peak when methane W ~ S passed
through the discharge. Trace quantities of higher molecular weight
hydrocarbons were also noted. The most significant features of
the methyl+chloride spectra (Figure lb) were the increaae in the
m/e - 2 (H2) peak, the appearance o$ the m/e = 30 (C2H6) peak and
the decrease in the m/e = 50 (CH3C1 ) peak when methyl chloride
was passed though the discharge. Further, HC1 was noted when
the liquid nitrogen cooled trap downstream from the discharge
was opened.
B. Solid Films
A solid film was observed to form in the discharge region
when either methane or methyl chloride was passed through the dis-
charge. Both films were characterized in several ways.
1. Elemental analysis
Empirical formulas for the solid films were established
by elemental analysis. The formulas for the solid films formed
form methane and methyl chloride were (CH1.5)x and (CHo.65Clo.045 1
respectively .
2. Infrared spectra
The infrared spectrum of the film obtained from methane
is shown in Figure 2 where a neat sample was used. Peaks ascribed
to the film were noted in the regipn of the following frequencies5
2900, 1700, 1450, 1380 and 880 cm . The assignments of these
freauencies follow those given by Jesch et al(2). The band at 880
cm hay be due to rocking vibrsfion of -CH in a multiple carbon
atom chain. The band at 1290 cm is characgeristic of -CH deforma-
3
tion. The band at 1450 cm is character’stic of - CH2 symmetric
scissors vibration. The band at 1700 cm-‘ was probably due gy
carbonyl formed on exposure of the film to air. The 2900 cm band
!

4
i
J
1
i
475
is due to C-H stretching vibration.
The infrared spectrum of the film obtained from methyl
chloride is shcyln in Figure 3 where a neat sample was used. The
low intensity peaks were a characteristic of the methyl chloride
films whether run as a neat sample or as a KBr pellet. Peaks
ascribed to the film were noted in the region-pf the following
frequencies: 2900, 1700, 1580, 1440 and 875 cm . The peak at
1380 cm observed in the methane film- s absent in the methyl
chloride film and a new peak at 1580 cm iis observed in tHe methyl
chloride film.
3. Electron spin resonance
The electron spin resonance spectra of the films pro-
duced from methane and methyl chloride are shown in Figures 4a
and 4b. A significantly larger free spin concentration on the
order of one thousand times greater was noted in the case of the
methyl chloride film. No fine structure was observed in the case
of the methyl chloride film. g-values for both films were
essentially identical to the value for pitch(g = 2.000).
4. Solubility studies
An extensive series of liquids were tested as possible
Solvents for the films prior to NMR measurements. Solubility of
the methane film was observed only in the case of hexamethyl-
phosphoamide and concd. H2S04.
IV. DISCUSSION
Investigations of chlorinated hydrocarbons in a microwave
discharge have not been reported in the literature previously.
Swift et a1 (3) however have investigated the effect of an rf
discharge on CC14 in the absence of a diluent. A number of
chlorinated gaseous products were reported in addition to a
chlorinated polymer. The present work indicated a limited number
of gaseous products. The results are consistent with the argument
that molecules in a xicrowave discharge are subjected to a
greater degree of fragmentation than in an rf discharge thereby
limiting both the number aild ;he complexity of gaseous products.
From the previous work of Vastola and Wightplan (1) the
following postulate could be advanced: if a parent hydrocarbon
has a hydrogen to carbon (H/C) ratio greater than about 1.6, a
hydrogen saturated solid film will be produced on passage of the
hydrocarbon through a microwave discharge in addition to hydrogen.
Conversely, if a parent hydrocarbon has a (H/C) ratio less than about
1.6 a hydrogen deficient film will be produced and no hydrogen
will be observed.
The present work is an attempt to test the validity of
this postulate if a functional group in introduced into the hydro-
carbon molecule. Methyl chloride was chosen since it represents
a straightfarward extension of the methane case. Methyl chloride
has a (H/C) ratio of 3 and if the C1 is neglected, the formation
of a hydrogen saturated film would be predicted. However, HC1
was observed as a rearrangement product. Hence the C1 can not
be neglected and yet assuming the limiting case of a 1:l corres-
pondence between CH3C1 and HC1, the (H/C) ratio of the remaining
fragment would be 2. Thus the formation of a hydrogen saturated
film would still be predicted but which in fact was not observed.

476
Instead a hydrogen deficient film was produced. The formation of
a hydrogen deficient film from a parent molecule which has a high
enough (H/C) ratio to form a hydrogen saturated film can be attributed
to two factors. In the first instance, C1 preferentially
appears in the gas phase as is indicated by the low percentage of
C1 in the film, the absence of a significant C-C1 absorption peak
in the infrared spectrum(Figure 3) and the presence of HC1 downstream
from the discharge. The extension of this work to other
functional groups is anticipated to determine if this is a general
scheme. In the second instance, the tendency to form hydrogen in
a microwave discharge is again noted in the case of methyl chloride
as in the case of methane(Pigures la and lb). Apparently for
hydrocarbons containing functional groups, the formation of hydrogen
is favored even at the expense of the formation of a hydrogen
deficient film. The formation of ethane observed as a product of
the methyl chloride discharge could be due to the recombination
of methyl radicals produced in the discharge.
Films formed in various types of electrical discharges have
not been extensively characterized. The recent work of Jesch et a1
(2) described the infrared analysis of a series of films produced
from hydrocarbons in a glow discharge. Direct comparison between
the present results and those of Jesch et a1 is not possible since
two different types of electrical discharges were used. The type
of discharge used dramatically alters the nature of both the gaseous
and solid rearrangement products as indicated above
The properties of the film produced from methyl chloride
are similar to those reported previously(1) for the hydrogen
deficient films formed from acetylene, benzene and naphthalene.
The color of the methyl chloride f ilm was dark (brownish-black)
characteristic of hydrogen deficient films in contrast to the light
yellow methane film characteristic pf hydrogen saturated films.
The high electron spin concentration of the methyl chloride film
ibad also been oberved for the hydrogen deficient film from acetylene.
The films produced from both methane and methyl chloride
appear to be highly cross-linked as evidenced by the negligible
solubility is liquids used as solvents for other polymeric systems.
The methyl chloride film contains a greater number of free electrons
than the methane film indicative of a significant number of unsaturated
valences in the methyl chloride film. The methyl chloride film
appears to be characterized by a greater degree of unsaturation
than the methane film. Highly unsaturated polymeric systems have
been found difficult to analyze by infrared spectroscopy* which
has also been noted in the present results. Definitive NMR work
would be helpful in elucidating the nature of these polymeric films.
* private communication with Dr. Vernon Bell(NASA - Langley Research
Center)
Acknowledgements - The authors would like to acknowledge the help of
Dr. R. E. Dessy in obtaining the ESR spectra and the experimental
assistance of Mr. R. 0. McGuffin‘.
REFERENCES
(1). F. J. Vastola and J. P. Wightman, J. Appl. Chem., 14, 69 (1964).
(2). K. Jesch, J. E. Bloor and P. L. Kronick, J. Poly. Sci. A-1, 4,
1487 (1966).
(3). E. Swift, ‘r., R. L. Sung, J. Doyle and J. K. Stille, J. Org.
Chem., a, 3114 (1965).

1
I
I
/ 10000
Y
1
-
I
before
di sc harge
I discharge
during
Fig.la. Selected
mass .spectra o
ih a microwave
2 15
MAS SIC HA RGE RATIO
eaks from Fig.1b. Selected peaks from
Pmethane mass spectra of methyl
discharge. chloride in a discharge.
n

i
478

L
t
479

480
FATTY ACIDS AND n-ALKANES IN GREEN
RIVER OIL SHALE: CHANGES WITH DEPTH
Dale L. Lawlor and W. E. Robinson
Laramie Petroleum Research Center, Laramie, Wyoming
INTRODUCTION
Maqy papers have been published relating fatty acids to n-alkanes in ancient sedi-
ments. Breger (1) postulated that n-alkanes may be formed in sediments by beta-
keto acid decarboxylation. Jurg and Eisma (5) demonstrated in the laborator) th6:
a homologous series of n-alkanes can be produced by heating the C22 fatty aci6 ir.
the presence of bentonite. Cooper and Bray (3 suggested that odd-carbon-numbered
n-alkanes and odd-carbon-numbered fatty acids may be produced from naturally occur-
ring even-carbon-numbered fatty acids by a free-radical decarboxylation mech-plsin
Mair (I) and Martin and coworkers (a) have discussed the relationship between fa-r,?
acids and n-alkanes in petroleum genesis.
If fatty acids are converted to n-alkanes in sediments, a distributional relation-
ship would likely be apparent between the two compound classes. Such a relationship
would be most relevant if both the acids and the alkanes were indigenous to each
sample taken from one geological formation. In this respect the Green River Forma-
tion is ideal for study of the relationship between fatty acids and n-alkanes. This
formation represents 6 million years of accumulation of organic debris from a highly
productive Eocene lake which existed approximately 50 million years ago. A 900-foot
core representing 3 million years of the Green River Formation was sampled and the
fatty acids were extracted and analyzed.
An earlier paper (5) presented a correlation between the carbon number distribution
of the fatty acids and the carbon number distribution of the n-alkanes in the
Mahogany zone portion of the Formation. The present paper tests the postulate that
n-alkanes are formed from fatty acids by maturation processes resulting from the
time differential existing between the bottom and the top of the core. The results
showed that both maturation and changes in organic source material explain the dif-
ferences in samples taken from several stratigraphic positions in the formation.
EXPERIMENTAL
Ten oil-shale samples were taken from the Green River Formation, nine from a 900-
foot core (Equity Oil Co., Sulfur Creek No. lo), and one from the Mahogany zone.
The stratigraphic positions and the oil yields of the 10 samples are presented in
table 1. The samples were the same as those used by Robinson and coworkers (2) in
a study of the distribution of n-alkanes at different stratigraphic positions within
the formation.
The samples, ground to pass a 100 mesh screen, were treated with 10 percent hydrochloric
acid to remove mineral carbonates and to convert salts of acids to free
acids. Two-hundred grams of each of the acid-treated samples were refluxed with 400
ml of 7 percent borontrifluoride in methyl alcohol for 6 hours. This reaction converted
free and esterified acids to methyl esters. The reaction mixture was cooled,
filtered, and washed with methyl alcohol until the washing was clear. The filtrate
was transferred to a separatory funnel, water was added, and the solution wae
extracted with successive portions of carbon tetrachloride until the solvent was
colorless. The carbon tetrachloride solution of the methyl esters was dried OV(?
night using anhydrous sodium sulfate. The fatty acid methyl esters were isolated
from the carbon tetrachloride solution by urea adduction.
I

TA?.LE 1, - StratigraphiL po:itio;. ar.d pyrolytic cil
yields of the oil-shale samples
Approximate pyrolytic
Samp le Stratigraphic position, oil yield, gallons per
Vamber feet from surface ton of shale
0
i
)
3
- I.
5
6
I
6
9
93 0
1036-105b
1056 - 1081
1152-11 ia
123b-12-1
1450-1L62
1597-1628
1668 -1696
1786 -1825
168G - 19 i-
Fatty acid methyl esters were further purified by thin layer chromatography using
silica gel The thin layer plate was developed with a mixture of n-hexane, ethyl
ether. acd acetic acid (9O/lO,'lj. The adsorbcnt zone where esters occur was scraped
from the plate and extracted with a solution of 10 percent methyl alcohol in carbon
tetrachloride The solvent was remoLed from the methyl esters by evaporation at
room temperature under a stream of nitrogeg" The carbon number distribution of these
isolated fatty acid methyl esters was determined 5y mass spectrometry using the
intensity parent peak as a measure of the quantity
Carboxyl aqd ester coqtents were determined for the 10 shale samples by analytical
procedurts deieloped by Fester and hsbixon (I).
RESirLTS AbD DISCJSSIOK
Thc fatty acid distribution with,depth (table 2), shows that the predominant acid at
various stratigraphic depths within the formation is not constant. The sample number,
percentage, acd chain length of the predominant acid of the samples are: No. 0, 16.2,
C30; NO. 1, 14.1, C24; NO. 2, 21.7, c28; NO. 3, 9.8, c28; NO. 4, 16.6, C24; NO. 5,
9.2, c28; No. 6, 12.8, C14; No. 7, 12.6, Clb; No. 8, 12.1, c28; and No. 9, 23.9, c28.
The C28 acid is the predominant acid in five of the samples, and is among the three
most abundant acids in all the other samples except samples 1 and 4 where it is pres-
ent in small amounts. The C20 acid content in samples 5 through 9, despite being
more abundant in nature than the C1g and e21 acids, is less than the C1g and C21
acids in these samples. The c18 acid is quite low in samples 1 and 2 relative to
the other eight samples. The C32 acid is present in large amounts only in sample
Yo. 0. There appears to be no consistent relationship between the acids of one Sam-
ple to that of another sample. This lack of relationship may bcdue to differences
in source material or environmental changes rather than maturation changes.
Significant differences are apparent in the .relative distribution of the fatty acids
and the n-alkanes from published data (Q), shown in figures 1 and 2, where the data
from four samples are plotted for comparative purposes. These particular samples
were chosen because they represent three stratigraphic levels below the Mahogany zone
sample approximately equi-distant from each other. If maturation of sediments causes
decarboxylation of fatty acids to form n-alkanes, a relationship between the distribulion.
of the two components would be expected. Except for sample No. 0, little.
similarity is evident. Since the distributions correlate poorly, either the original
postulate is 11r.true and n-alkanes are not produced from fatty acids having one more
60
35
34
42
it8
39
35
37
28
20

482
TA5L,E 2. - Carbon number distributions of fatty acids
Fcttv acids, wt percent of total
Carbon Sample number
Numb e r 0 1 2 3 h 5 6 7 8 9
14
15
16
17
ia
19
20
21
22
23 '
24
25
26
27
28
29
30
31
32
5.1 7.1
3.0 3.4
2.7 7.1
2.0 4.0
3.4 2.6
2.4 3.4
2.7 8.2
2.4 7.1
7.4 8.8
4.7 4.7
7.1 14.1
3.7 7.6
12.1 11.8
2.7 2.6
11.1 4.2
2.7 1.3
16.2 1.3
1.7 0.4
7.1 0.4
2.8
2.8
2.4
1.9
2.1
1.8
1.9
2.2
3 .o
4.4
6.2
5.0
13.3
8.2
21.7
5.1
8.6
2.5
2.9
6.3 :.3 8.0 12.8
6.3 2.8 7.2 2.1
5.6 4.0 5.8 6.4
4.9 2.9 5.1 5.9
4.2 4.2 5.i 8.0
3.5 3.8 4.0 4.3
2.8
2.8
L.5
5.6
2.9
3.6
4.2
4.8
4.9 9:; 5.8 11.6
5.6 i2.5 5.8 3.8
4.2 16.6 3.3 8.0
3.5 8.3 1.5 3.8
8.4 11.8 6.5 4.9
6.3 3.7 8.0 2.7
9.8 4.0 9.2 9.8
4.2 1.2 4.3 1.5
7.7 1.1 5.8 2.5
4.2 0.3 4.0 0.8
4.9 0.5 3.3 0.9
12 .,6
6.3
7.8
4.3
7.4
3.6
2.9
3.6
8.3
4.7
6.3
3.4
5.9
4.9
8.5
2.9
3.2
1.6
1.6.
11.4 6.5
8.1 3.7
9.4 5.4
2.4 2.0
7.4 6.1
2.0 2.0
2.0 1.3
2.0 1.7
10.7 13.0
3.4 3.3
8.7 12.0
2.7 2.6
4.1 5.2
3.4 3.7
12.1 23.9
2.7 2.4
2.7 2.8
2.0 1.3
2.0 1.1
carbon atom or other factors are more significant. One factor to consider is that
the fatty acids recovered from the lower portions of the core may not be representa-
tive of the acids originally present since they may be residual acids from selective
maturation.
Groups of even numbered acids vary in abundance as shown graphically in figures 1
and 2. Sample No. 0 has predominant acid peaks between c26 and C30, sample No. 4,
between C22 and C26, sample No. 6, between C22 acd C28, and sample No. 9, between
C22 and c28. The latter two samples show the greatest overall similarity.
The carboxyl content of the shale generally decreases with depth, table 3, beginning
with sample No. 0 having 23.6 mg of carboxyl per gm of organic carbon and ending
with sample No. 9 having 4.8 mg of carboxyl per gm of organic carbon. This general
decrease in carboxyl content with depth, and a corresponding increase in the n-alkane
content, as shown by Robinson and coworkers (9, implies that the fatty acids may
have been converted to n-alkanes by decarboxylation.
The ester content of the samples, table 3, did not show a trend; this is in contrast
to the trend shown by the acids. Random distributions found for the ester samples
reflect more accurately changes in organic source material than do the acid
distributions.
Maturation changes are evident in the decreasing amount of carboxyl content with
depth, the increasing amount of n-alkane content with depth, and the decreasing
ratio of odd to even-carbon-numbered n-alkanes. Source material or environmental
changes are evident in the variations in the amount and carbon number of the domi-
nant fatty acids, and n-alkanes and the random distribution of ester content with
depth.

1.
2.
3.
4.
5.
TABLE 3. - Carbowl and esler contents of the
10. shale samles
Samp 1 e Carboxyl, Ester,
Number mgf rn carbon mgfgm carbodf
0 23.6 36.6
1 18.3 22.2
2 10.1 33.9
.?I! -- --
4 3.8 26l.7
5 6 7 1; .q
6 ~4.0 37.1
7 4.7 26.0
8 4.9 42.9
9 k.8 29.3
- 1' Ca,culated as COOH
?/ 1ns.ifficient sample.
CONCLUSIONS
73e postulate that fatty acids from the organic debris in the Green River Formation
may form n-alkanes of one less carbon atom was found to be subject to question.
Relatable fatty acid and n-alkane distributions were shown to be preeent in the
younger portions of the Formation but the older portions showed little relation.
batbration and organic source material differences may account for the obeenved dis-
tri5ut;ons of fatty acids in sections of Green River Formation oil shale. Evidence
of matLration is found in the decreasing amount of odd-carbon-numbered n-alkanes and
in the decreasing amount of carboxyl content with depth. Evidence of organic source
material differences is found in the inconsistent distribution of the predominant
fatty acids from sample to sample.
The ester content is not related to the free-acid content, indicating different
formative histories.
REFERENCES
Breger, I. A. Diagenesis of Metabolites and a Discussion of the Origin of
Petroleum Hydrocarbons. Geochim. et Cosmochim. Acta, v. 19, 1961, pp. 297-308.
Cooper, J. E., and E. E. Bray. A Postulated Role of Fatty Acids in Petroleum
Formation. Geochim. et Cosmochim. Acta, v. 27, 1963, pp. 1113-1127.
Cumins, 3. J., and W. E. Robinson. Normal and Isoprenoid Hydrocarbons Isolated
from Oil-Shale Bitumen. J. Chem. Eng. Data, v. 9, 1964, Pp. 304-301.
Fester, J. I., and W. E. Robinson. Oxygen Functional Groups in Green River Oil
Shale and Trona Acids. Ch. in "Advances in Cheeietry Series." American Chemical
Society, Washington, D.C. (1966), pp. 22-31.
.'urg, J. W ., and E. Eisma. Petroleum Nydrocarbons: Generation from Fatty Acide.
Science, v 144, 196i, pp. 1451-1152.

6.
7.
a.
9.
484
Lawlor, D. L., and W. E. Robinson. Division of Petroleum Chemistry, Am. Chem.
SOC., Detroit, Mich., April 1965, Petroleum Division General Papers No. 1, 1965,
pp. 5-9.
Mair, B. J. Terpenoids, Fatty Acids and Alcohols as Source Materials for
Petroleum Hydrocarbons. Geochim. et Cosmochim. Acta, v. 28, 1964, pp. 1303-1321
Martin, R. L., J. C. Winters, and J. A. Williams. Distributions1 of n-Paraffins
in Crude Oils and Their Implication to Origin of Petroleum. Nature, v. 199,
1963, pp. 110-114.
Robinson, W. E., J. J. Cumins, and G. U. Dinneen. Changes in Green River Oil
Shale Paraffins with Depth. Geochim. et Cosmochim. Acta, v. 29, 1965, pp.
249 - 258.
h
1
c
/
‘I

i
I
I
,
r
t
\ ..
4
3
I I I I I I 1 . 1 I
13 17 21 25 29 13 17 21 25 29
ALKANE CARBON NUMBER
I
'1 I
F&ure 1. Caupiwison between fatty acld a d n-allme
distributions, Sample Uos. 0 and 4
.. .

' I ' Sample No. 6 '
12
IO
8
6
z 4
z
4
52
I 1 I I
4
0/4 I: i2 /6 . i0 Ib I\ ;2 26 '
v)
n
$14
ACID CARBON NUMBER
t24.7
Sample No.6
n-AI kaner
Sample No. 9
n-AI kanes
13 17 21 25 29 13 17 21 25 29
ALKANE CARBON NUMBER '
Figure 2. m i s o n between fatty acid and n-alhue
diatributions, sample Nos. 6 and 9

487
ELECTRICAL PROPERTIES OF IODINE COXPLEXES OF ASPHALTENES
Gustave A. Sill and Teh Fu Yen
Mellon Institute I
Pittsburgh, Pennsylvania 152 13
INTRODUCTION
Polynuclear aromatic hydrocarbons generally form charge-transfer
complexes with halogens. Some of the fused aromatic hydrocarbons, e.g.,
perylene, violanthrene, yield solid complexes exhibiting extremely good
semiconduction (l,2,3) while others, e.g., coronene, show only fair to
poor semiconduction (4).
A number of charge-transfer complexes of aromatic
containing polymers have been investigated for possible differences in
electrical properties (5,6,30).
A polymeric dielectric may be converted to a polymeric semi-
conductor by increasing the aromaticity of the insulator, followed by
complex formation with a halogen (7,8). The increase in aromaticity can
be effected by radiation--e.g., cyclization of polyethylene and followed
by dehydrogenation (7); or by heat--e. g., pyrolysis or graphitization to
a pyropolymer (8).
lhe resulting products when treated with iodine exhibit
a wide range of interesting electrical properties.
From a structural standpoint asphaltenes (9) are considered to
consist of two-dimensioned fabrics of condensed aromatic rings, intermingled
with short aliphatic chains and fused naphthenic ring systems (10).
diffraction (11,12) and ESR investigations (13) have indicated that these
aromatic oyotemo tend to form otacko of graphite-like lryrro rurroundrd
by a disorganized zig-zag chain structure of saturated carbons.
they may
be considered as a highly associated "multipolymer" (14), the
X-ray
Morphologically

.
488
uiolecular weight of which can vary from a few thousand (unit or particle
weight) to a few million (micelle weight) (IS).
The aromatic centers,
roughly 15; in diameter, are considcred to be pericondensed (~3,iB).
has been demonstrated that asphaltics can form charge-transfer complexes,
due to the presence of such aromatic systems (20).
Most polynuclear aromatic compounds form w ell defined crystals,
the iodine complexes of which are stable and stoichiometric in composition.
Asphaltics are mesomorphic (17) owing to the raridom distribution and iso-
tropic orientation of the structural units, and it is to be anticipated,
therefore, that the conduction mechanism will be different from that of
the crystalline environment (due to diffusion and phonon processes) (16). 1
\.
Since there is a difference in the conduction mechanism between crystalline
and amorphous aromatics, one would like to know whether the mesomorphic
nature of asphaltics would retard or inhibit the conductivity, and if so )I
to what extent. I
The aims of this research were two-fold. The first was simply
to observe where asphaltenes fall in the conductivity range and to
determine the extent to which conductivity can be enhanced by iodine com-
plex formation. The second was a more general study of the electrical
properties of asphaltics as another approach to a better understanding
of their structure. To the authors' best knowledge, there is no published
work on the electrical characteristics of these materials. Iodine may
be visualized as a tracer or indicator for condensed aromatic systems,
even when buried in a matrix of paraffinic or cycloparaffinic material.
It was thought, therefore, that iodine complex formation and its effect
It
I
r

489
on the ovcrall electrical properties of the asphaltene might yield inde-
pendent information of the size anti distribution of the aromatic centers
I
in these multipolymers.
Res 1 stance Eka sur cment s
ESPERIPfEhTAL
A General Radio type-1230b electrometer was used for specimens
with resistance less than 1012 ohms (ambient ,temperature), and a Cary
model 31 electrometer was used for specimens of higher resistance. In
each case a glass vessel equipped with a ball joint and appropriate
e1ec:rostatic snielding was coupled to the head of the electrometer (Fig. 1).
specimen (approximately 1 x 0.5 x 0.1 cm) was pelleted with a Beckman KBr
press at 7.09 x lo3 kg/cm2 between two pieces of 52 mesh platinum screen.
The pellet was degassed for eight hours and the electrical measurements
made in a vacuum of 5 x Torr. Using the high resistance leak method,
a standard resistor served as calibrating reference (21); data were taken
under conditions of both falling and rising temperature, a minimum of 30
minutes being allowed for equilibration at 10" levels. Upon completion
of the temperature dependence measurements, the physical dimensions of
the specimen block were obtained with a travelling microscope (1OX) with
an x,y micrometer attachment (0.0001 cm precision).
Preparation of Sample
The asphaltene sample was prepared by our standard procedure (9).
Two native asphaltenes were investigated, one from the Boscan crude oil
from Venezuela (Sample VY), the other from the Baxterville crude from
Each

Xississippi (Sample GS).
Stock solutions in benzene of iodine and of the
individual asphaltenes were made up in fixed concentration and samples
I
of varied composition prepared by mixing appropriate quantities of these
stock solutions at room temperature and lyophylizing at reduced pressure
to yield powdered solids with a homogeneous iodine distribution. These
samTles were analyzed before and after the electrical measurements for
$1 by ignition in oxygen, reduction with hydrazine sulfate, and poteniometric ,
titration of the resulting iodide with &NO3 using a Beckman model K
automatic titrator. Usually there was no observable loss of free iodine
during electrical measurements; lO-l5$ loss of iodine was found after
degassing. The iodine values used in the present work are those values
obtained after completion of the electrical measurements for the entire
specimen.
Treatment of Data
The resistance values along with the corresponding temperature
data and dimensions of the specimen were key punched on IEiM cards and
evaluated on an IBM 7090-1401 digital computer system. Given A, the area
of the cross section, and L, the Length of the specimen, the resistivity,
p, can be evaluated from the resistance, R, as follows:
p = AR/L. The
temperature dependence of the resistivity is then evaluated by the relationship:
4
PT = 00 exp (+kT), i
where k is Boltzman's constant, E is the energy gap in eV and po is the
resistivity extrapolated to ,$ = 0.
By use of a California Computer Products %
30-in. plotter (300 steps; l/5OO-in. per step) the temperature dependence
data were fitted to straight lines (Fig. 2) as given by the equation
\
<
\
\.
V
\
4

3'
I
/
3
2 b
V
t
491
log p = log pJ-€E/(2kT In 10)
From the digital output, p250c and E can be obtained. The applied voltage
I
was limited to values under LO V; in this region Ohm's law was followed.
The temperature range examined was from ambient to 90°C.
There is error involved in any single measurement of resistance,
owing to systematic errors in the electrometer; errors also enter in the
measurement of the dimensions of the specimen. That the results were not
influenced by such systematic errors is evident from the two sets of data
for two different preparations of an asphaltene (VY)-iodine complex, as
given in Table I. The uncertainties in the per cent iodine and sample
size may be judged from the variations in the independent measurements.
Despite these variations, the resistivity at 2 5 O C and the energy gap are
within ca. 5% of the mean values.
Infrared Analysis
Differential IR spectra were obtained from a scan of an iodine-
containing asphaltene versus a reference asphaltene at equal asphaltene
concentration in CSg (the iodine-containing sample is normalized to lo@$
asphaltene for purposes of comparison) using a Beckman IR-12 instrument.
A control scan of asphaltene in CS2 solution against itself also was made
for each sample.
X-ray Diffraction
A Norelco x-ray diffractometer equipped with a CuxGL radiation
source and a geiger tube detector was used to study the asphaltene-iodine
system. In order to record the shift of the d-spacing due only to change
in mass absorption coefficients, adamantane was added to an asphaltene-
iodine complex (24% I) and to the original asphaltene.
Strong (111) and
(260) reflections due to the adamantane mixed with the VY asphaltene were

492
found at 5.7 and 4.9& for the W asphaltene-iodine complex shifts were
observed to 5.5 and 4.7A, respectively. The spacing is reproducible to
I
9. ai.
Eleztron S?in Resonance
ESR spectra were taken with a Varian V-4502 x-brand EPR spectrometer
system in conjunction with a 12-inch magnet and a "Fieldial."
intenslty observed was used as a guide for the spin concentration of the
asphaltene (VY)-iodine complexes and native asphaltene (VY) (13).
RESULTS
The relative
All asphaltene-iodine samples studied gave repeatable linear
relations in the temperature range investigated as shown in Fig. 2.
is no significant deviation from Ohm's law through the range 2.5 to 97 V
as indicated by Fig. 3 in the temperature interval 313" to 372"K.
insulator range.
There
The native asphaltenes (Points 1, Fig. 4) fall generally in the
Upon the addition of iodine the resistivity f;Lls, 5 to
- b, then increases sharply, to c, and finally drops, 5 to a, as iodine
content rises. Both complexes appear to yield curves of similar shape.
The gap energy values for the asphaltene-iodine complexes are plotted
versus their iodine contents (Fig. 5). . The smallest energy gap measured
in each case was -Q.5eVf but the absolute minimum is uncertain, These
minima (points b) correspond to the sharp transitions of resistivity shown
in Fig. 4..
Scanning in the far IR region revealed no C-I stretching frequencies
for those iodine complexes for which 0 was determined. However, the dif-
ferential IR measured in the 7OO-l2OO cm-1 did show an additional band at

i'
I
Y
'
493
..
10E.G ~r,~-l {i;ig. 6) , and freshly ;,rc,,ar'cd asi'i~alecl-ic-iodine cosplexes in
CSz also csiiibited an enhanccmcnr in absorprion 'in the regiofi of lOC0 to
11 jO cn-1, w:IicIi is generally ascl-iilcu to complex iormation.
Tile x-ray spectrograms in rile region 20 = 2 - 42" of the asphaltene'
(mi)-iodine 'systca yielded in aiiior;),ious pattern with broad halos as shown
. .
in Xg. 7. For samples with an iodinc content of less tian 5 per cent,
the j. ji band.stil1 appeared as ii siioulder.
was fornad at a;oound 8.7a.
In thc meantime a new band
AT bigher'iodine conhenis (>IO$), the 3.5i
band. has apparently disappeared and the 8.7x band became clearly visible.
".
liisrc is a general overall decrease of total intensity as $1 increases
since the iodine itself absorbs an increasingly large fraction of the
.. .
diffracted x-rays. In the present case, the noise Level coupled with the
intensity reduction has made the disappearance of the 3.51 shoulder dif-
ficult KO detect.
In gcnera1,the . . ESR,spectra obtained froqthe asphaltene-iodine
samples indicated an incre,ase in free radical concentration with increase
-in iodine content. The increase in relacive intensity for an asphaltene
(Wj-iodine complex ,.containing 2@ I over the corresponding native asphal-
. .
tenes is about 3.8 fold.
DISCUSSIOX'
In all the samples studied, including both.native asphalcenes
and their iodine. adducts, a negative temperature coefficient of resistivity
was obtained.
by the plots shown.in Fig. 2:
The linearity of log D vs. reciprocal T .is substantiated
The resistivity is inversely related to
the concentration of the charge carriers (holes and electrons), but the
fact that the number of charge-carriers increases exponentially with tempera-
ture does not enable a choice to be made between electronic or ionic conduction
"

494
mechanisms. However, the voltade dependence for the current at a relatively
low electric field is linear, indicating that Ohm's law is valid (Fig. 3).
I
This adherence to Ohm's law supports the belief that the conduction is
electronic (19).
The fact that the asphaltene-iodine sample exhibits (Fig. 6)
strong enhancement of absorption near 1080 crn'l suggests that an iodine
molecule forms a donor-acceptor complex with the aromatic portion of the
asphaltenes. It is known that the iodine molecule forms such a charge-
transfer complex with benzene and other alkylated benzenes, and that
these complexes, in general, exhibit bands from 9 9 cm'l to 1200 cm'l
(24,25).
Arguing from analogy, it is plausible that the complex assumes
the axial model (model A of Mulliken) while the acceptor molecule is sitting
perpendicular to the plane. of the aromatics (26).
Further, our failure
to' locate any C-I stretching frequencies in the 400-600 cm-1 region sup-
ports the view that iodination of the asphaltene samples did not occur'.
crystalline.
Xost charge-transfer complexes of iodine with aromatics are
An exception to this is the violanthrene-iodine system,
which x-ray diffraction indicates to be amorphous (2).
X-ray results also indic'ate a low degree of order for the asphaltene-
iodine Complexes. If the acceptors (I2) are homogeneously distributed in
the host matrix (asphaltene), these systems may be considered analogous
to the impurity or valence-controlled semiconductor systems. Disappearance
of the 3.5i spacing of the (002) band means that the layered structure of
asphaltene must have been altered (Fig. 7).
The 4.68, y-band, due to the
saturated carbon in the structure, did not change in the complexing process.

d
:- /
I
i'
I
495
:"ne ~CIJ hznd A: S.?h thcn nay bo iuc to the cx;>ansion of the zironatic
intcrplsnzr. distazia to aliow for coir,plexing by the iodine no;ecule.
0
thc case of the ;>eryicnc-iodinc systcns, Liie spacing found at 10.7A was
interpreced (3) as the distance bctwecn perylene moleculcs when iodine
moiccnlcs were sandwiched between tiic aromatic layers. The present ob-
served value of &.?A can be viewed as the sua of the intcrplanar distance
(5.5.:) and the iodine length '(4 x l.j$).
linked layers resembles that oi an intercalation compound of graphite
(Fig. 6).
Tile picture of these inter-
. .
Beferring again to Fig. 4, the room temperature resistivities for
the two nztive asphaltenes are seen to be in tne insulator range (>lo14 Ohm).
Upon addition of the iodine, the resistivity decreases about six decades
or a niLlion fold. This is essentially the same behavior as that observed
for polymeric charge-transfer complexes such as poly(viny1pyridium TCNQ)
ar.6 its derivatives.
These polymeric salts are dependent upon the TCNQ con-
centration wnicn at best increases the conductivity six decades. Here
!lave to point out that it is not easy to prepare a polymeric charge-
rransfer complex with good semiconduction.
In
Slough (5) nade a number of
polymeric complexes.from arosatic-containing polymers, with acceptors such
GS cetracyanoechylene, chloranil, etc., and found the conductivities of
these conplexes were not measurably higher than those of the original poly-
meric donors.
These mesomorphic materials may lack the order of the K-
systems needed to open a path for charge carriers.
More likely, the
aroGtic systems are too small to form stable charge-transfer complexes.

For pure polynuclcar arom.i;ic hydrocarbons, donor-acceptor com-
Irlescs, cxpccially the iodine coinpicscs (2,3), exhibit increases in con-
I ,
ductivity of 12 to 16 decadcs when compared to the parent hydrocarbon.
The enhancement in conduction by the addition of iodine can be illustrated
by comparison of the complexes with a valence controlled semiconductor, a typical
esample being nickel (11) oxide doped with lithium oxide (22).
In Fig. 9 a com-
parison is made between this system and the aromatic-iodine complexes. The
trends in resistivity with the concentration of impurity are quite similar.
- Asphaltenes contain fused ring aromatics, the peripheral hydrogens
of which are substituted heavily by short chain alkyl groups (23).
to the relatively large porportion of methyl groups (20$) and the large
Owing
average layer diameters (La - 1$), the asphaltic molecule can be viewed
as a typical aromatic donor ( D); halogens such as iodine can behave as an
acceptor (A). Through charge-transfer by overlapping of the molecular I
!
orbitals of the two moieties, the dative structure (D'A') should result
in which asphaltene is the positive ion. \ ,
Fig. 4 clearly indicates that the increase of conductivity follows
two different paths, the first 5 to b, is the path followed for small in- <
crements of iodine, terminating at k with fixed composition (VY, 15.5%;
GS, lo.&); the other, c to rl, is for higher percentages of iodine.
1
Line & represents the transition state. The curve &may be extrapolated 1
to a resistivity value of 5.8 x 106 ohm-cm, corresponding to that for pure
12 (27)'. The same results were found (23) for violanthrene-iodine system with
a minimum corresponding to a 2:l iodine-violanthrene molar ratio for the
complex. Apparently in Fig. 4 corresponds to a stoichiometric ratio of
~
I
\ ,
\
\
\
a
7
1
a
\

497
a stable complex where conductivity is at a maximum.
It is assumed that
for iodinc contents less than that corresponding to the minimum the amount is
I
insuiiisicnt to form the complex. At tile resistivity is highly sensitive
to the number of iodinc molecules. When c is passed, excess iodine acts
(is (in impurity in the stoichiornctric complex.
reflecccd by the energy gap plot in Fig. 5. The same fixed composition
(VY, 15.5';;; GS, 10.W;) is obtained in cither plot.
the energy gap value is ca. 0.5cV suggesting favorable condicions for
conduction.
The transition at a is also
At these minima,
A number of aromatic-iodine complexes have been reported (2,3)
and from their phase diagrams (either temperature or density vs. mole
per cent of iodine) the complexes are found to be stoichiomecric (29).
For example perylene-iodine can have 2:3 or l:3; pyranthrene-iodine i s
1:2; violanthrene-iodine is 1:2; pyrene-iodine is 1:2. In all cases for
peri-type aromatics the ratio of 12 to aromatic is higher than unity.
Since the diameter of the aromatic system in asphaltics falls in the
range 8-l$ (12), the system would be comparable in size to violanthrene
or perylene.
Assuming the composition at the transition '(q 15.5%; GS, lo.@)
is stoichiometric, then for any given ratio of aromatic and iodine,
the molecular weight of the asphaltene can be calculated. We have taken
the liberty of calculating this weight for W and GS asphaltenes based
on sample ratios of 1a:asphaltene of 2:1, 3:2 and 1:l. Since all layers
contain the aromatic moieties and the sample is free of wax contamination,
the molecular weight obtained is that of the unit sheet weight (weight of an
average sheet containing both aromatic and saturated carbon atoms). These

-
'-'. . ' . .
values are listed in Table 11. Next, provided aromaticity is also known
(fa for W, 0.35; for GS, O.5l), the disk weight (weight of aromatic
carbon atoms in a single sheet) and the layer diameter also dan'be approxi-
mated. These values are also listed in Table 11. Experimental values
for the W and GS asphaltenes from a previous paper (15) are included: It i
is of interest that the unit weight values obtained from GPC, mass spectrometry,
x-ray diffraction and the electron microscopic measurements agree in general
magnitude with the weights obtained by the present method.
Deviations of the
disk weight of GS calculated from resistivity from that obtained by mass
spectrometry may by due to the polydispersity of the GS asphaltene (e.g.,
Mw/MN for VY is 1.27; for GS, 1.74 (15)).
From Table 11, the asphaltene-iodine complexes formed appear to
correspond to an 12:asphaltene ratio of about l.5:l. This composition is
J
shown in model A of Fig. 8.
Actually aromatic disks of the size present
in asphaltenes should be able to accomodate more than one molecule of iodine.
Finally the increase in the free spins, as demonstrated by the
EPR spectra, may be indicative of an increase in carrier concentration (2).
The nature of these charge carriers will have a strong bearing on the
conduction mechanism and will be the subject of a separate investigation. . <
ACKNOWLEDGMENT
The authors thank Mr. Timothy R. Drury, Mellon Institute, Research
Services, for his technical assistance; and Mrs. Marianne P. Hoy and
Dr. Sidney S. Pollack, Mellon Institute, Research Services for the x-ray
diffraction work.
This work was sponeored by Gulf Research & Development Company
as part of the research program of the Multiple Fellowship on Petroleum.
. .
, . I
.. .
.. . .
. .
.. .*. .i
. .
\
i
\\
\
i
7
(
t

499
(it Lnk>;t,icC,i: ii., ani k:taiiiatu, ii., Solid state i'iiysics, 2, ~3 (1561).
(2) Schida, T., and Akclmatu, li., liull. Ciicm. SOC. (Japan), 2, 921 (1562).
(;) ~c.li& , T., 9nd .q;aimtu, li., BULL. Clicm. ~oc. (Japan), 2, 1015 (1.961).
(1;) Lab~s: X. X., Sciir, R. A., and Bosc, hl., "Yrocccdings of the Princeton
';.;ivcrsicy Conr'crencc on Semiconductors in Molecular Solids,"
?ri;i$.ccon, S. J., 1960.
(j) Slau:.;ii, Id., Tyans. Faradzy SOC., 2, 2360 (1362).
(6) Baith, S. .4., Vannikov, A. V., Grishina, A. D., and Sizhnil, S. V.,
!). 'F.. \
(9) The asphzlccne fraction of a crude oil or refinery residue is a brown-
to-black, amorphous, powdery,. solid material, insoluble in n-
;>i'i;tAI>e but soluble in bcnzene; see Ref. 11 for sample prepara-
tion.
(10) Yen, T. F., and Erdman, J. G., A. C. S., Div. Petroleum Chem.,
PreTrints, 1, No. 1, 5 (1962).
(11) Yen, T. I?., Erdman, J. G., and Pollack, S. S., Anal. Cnem., z, 1587
(1561).
I .
,2) Yea, T. F., and Erdnan, J. G., in "Encyclopedia of X-rays and Gammarays"
(G. L. ' Clark, Ed. ), Reinhold Publishing Co.,
s. Y., 1963, p. 65.
New York,
(l>) Ycn, T. F., Erdman, J. G., and Saraceno, A. J., Anal. Chem., 3,
694 (1952).
(14) The WOYL "multipolper" refers: to a polymer with a very large number
ijL different buildin; blocks, say ca. 100. Simpler structures
can Dc referred to by more specific terms, as for example,
"copolymer" (2 blocks) or "terpolymer" (3 blocks).

500
(1;) Ten, T. 17.: a;1d,Dic;

'/
925°C. x lO"2
Sn.:>Lc SL>." "Iodine ..\ (cm) I. (cm) OLlrn-cn E (CVY
23 i7.1 1.252, 0.0994 1.75c 1.742
/ 25 16.6 1.266.L 0.1053 1.61, 1.825
27 17.2 1.2592 0.1192 1.773 2. 007
(a) Sane namers used in curves 95 Fig. 4; Sample Nos. 2j, 25, and 27 are from
\ one ?reparatLon; Sample Nos. 24, 28, and 29 are fros an-
d other preparation.
i
I
c
1'1
\I

- ~
Resistivity Czlculation
2: 1
j:2
1: i
x-ray DL ffrsct ionh
(La)
502
27?Od k570 %9e 2330 1k.6f 22.6
2080 34jO 727 1750 12.6 19.5,
1360 2290 484 1170 10.3 16.0
- - - 11.9 17.0
E 1 ec t ron Xi cros cope" J
.-
(Particle weight) 3440 LO30 - - - -
(a) Weight 0; a single sheet containing both aromatic and saturated carbon atoms
(see 2. P. Dickie and T. F. Yen, A. C. S. , Div. Petroleum
C'nem. , Preprints, Miami meeting, April, 1967).
(b) Weight of aromatic carbon atoms in a single sheet.
(c) Diameter of aronatic cluster; see T. F. Yen, J. G. Erdman, and S. S. Pollack,
Anal. Chem. , u, 1587 (1961).
(d) Calculated based on 254t'lR (LOO-t; where R is the ratio of 12 to asphaltene
ana t is the %I corresponding to the transition in resistivity and
gap energy.
(e) Calculated from fa x (unit weight).
(f) La = (2.62 CA);/', CA from disk vcight without contribution of hydrogens.
(g) From pel permestion chromatography data on the native asphaltene, number average
rdiecular weight. 1,
(h) Expcrio.cn:ai dct; from J. P. Dickie and T. F. Yen, A. C. S., Div. Petrolen=
C:ics. , Preprints, Miami rr,eet;ng, April, 1967.
(i) J. P. DLciiic ar.d T. F. Yen. A. C. S., Div. Petroleum Cilem., Preprints,
11, 39 (1%6).
1
?
L
-.
\
,
,

503

504
o '
C
0s 2
0
v
x
i
8
2
0
!
P
-
a,
0
C
0
in
.-
In
a,
cc
u-
0
al
0
C
a,
D
C
a,
9
n

505
up”
/
U
d
L
.-

506
i

507
c

d 601
L
P
LL

I'
i
i
C
.. .
i?
2
s

I I I I I I I I
m W N
W N 0 (D
?
0 2 0 2 0 0 0 0
D
N
IO
,

, 513
THE ELECTRON SPIN RESONANCE SPECTRA OF CELLULOSE
CHARS TREATED WITH HALOGENS*
Evidence for Donor-Acceptor Complexes in Coals and Chars
R.M. Elofson and K.F. Schulz
Research Counci I of AI berta
Edmonton, Alberta
INTRODUCTION
Coals and low- and medium-temperature chars have been shown to be paramagnetic
by bulk magnetic susceptibility measurementsl,2fr3 and by electron spin resonance studies4r5t6.
If due care is taken to eliminate errors, reasonable agreement in the estimate of the number of
unpaired electrons present can be obtained'. While there is agreement on the existence of
paramagnetic centres in these materials, the origin and nature of these free radicals is not
understood8,"10,'?,12. It is the purpose of this article to present new evidence to suggest that
these unpaired electrons arise from donor-acceptor forces present in these materials.
One way of studying the nature of free radicols is by observing the behavior of the
electron spin resonance signals when the substrate is treated with adsorbed gases and other reagents.
Early workers in the field found that poramagnetic gases such as oxygen affected only the spin-
lattice relaxation times of chars prepared at from 300" to 5OOOC. For chars prepared in the 550'
to 800" temperature range, however, reversible line broadening occurred with an apparent
decrease in free radical concentration at higher oxygen pressures. Removal of oxygen restored
the original signal, which indicated that the phenomenon was due to physical and not chemical
processes. Halogens were at first found to have no effect on coals and chars and the oxygen
effect was attributed to spin broadening due to the paramagnetism of the oxygen molecules.
Similar results to oxygen were obtained with nitric oxide.
In 1963 Wynne-Jone~~~ and co-workers found that e.s.r. signals in chars obtained
by heating polyvinylidene chloride (saran) were affected by adsorption of halogens. These
workers noted that IBr, ICI, I, Br, and C12 caused partial or complete reduction in spin
concentration at constant line width. The original signal was not restored by outgassing in vacuo
at room temperature but could be partially restored by reheating the specimen at 100°C in vacuo.
Since, as mentioned above, no halogen effect had been obtained on coals or more common chars,
these authors attributed the susceptibility to halogens of these saran chars to a high degree of
porosity. Reinvestigation of the interaction of a series of cellulose chars with iodine, bromine
and hydrogen iodide in this laboratory has shown pronounced effects on the electron spin resonance
of some of these materials.
EXPERIMENTAL
&llulose chars were prepared by fint charring cellulose powder (Whatman Cellulose
Powder, Standard Grade) in an autoclave at 195"-2OO0C under nitrogen. Subsequently this char
* Contribution No. 360 from the Research Council of Alberta, Edmonton, Alberta.

514
was heated under nitrogen in a Vycor tube to the desired temperature and held at temperature for
one hour. Aliquots of the chars were then treated with a lawe.excess of 0.1 M solutions of iodine
or bromine in carbon tetrachloride. After standing in the solutions for twenty-four hours, the
samples were washed once with carbon tetrachloride and dried. Samples for e.s.r. measurements
were ploced in 3 mm. 0.d. glass tubes and evacuated.
Samples to be treated with hydrogen iodide were placed in open 3 mm. 0.d. tubes
placed in the cavity of the spectrometer and purged with purified nitrogen. Gaious hydrogen
iodide (Matheson) was passed directly over the sample in the cavity. Excess hydrogen iodide was
removed after treatment by subsequent purging with nitrogen.
Electron spin resonance measurements were performed with a Varian Model 4500
electron spin resonance spectrometer fitted with a TE102 cavity and a 100 Kc modulation attachment.
Saturation phenomena were avoided by working at 10 db attenuation and check for saturation were
made. Line widths ond spin concentrations were measured from first derivative cutves, generally
by direct inspection but in the case of very wide and very narrow signals resort was made to
graphical integration. A standard 500" cellulose char, used for calibration, was standardized
against DPPH, CuSO, and a Varian Standard char. The amount of sample was always less than
10 mg. and asymmetric line shapes were not observed, indicating the absence of serious skin effects.
RESULTS
In Figures 1 and 2 are summarized the results of iodine and bromine treatment on line
width and spin concentration as measured at room temperature. The untreated chars prepared at
625" and 650"show a marked decrease in line width as compared with those prepared at higher
or lower temperature -an effect attributed to exchange4. It is seen that both iodine and bromine
remove this intense narrowing in the 625" and 650" chars. Neither has o significant effect on the
300" to 500" chars but both broaden further the signal from the 700" char. The broadening due to
bromine i s not as great as that due to iodine and for the 600" char, in particular, the broadening
effect of the bromine is minimal. The effect on spin concentration shows that line broadening is
accompanied by a considerable drop in concentration in both the bromine and iodine treated chars.
Measurements made at 78°K showed that whereas the signals for the 300" to 500" chars
were broadened somewhot, those from the iodine- and bromine-treated chars in the 600" to 700"
range were narrowed. Washing the iodine-treated chars with cold 0.1 M sodium thiosulfate
solution restored the original narrow signals. This showed that no chemical reaction had occurred
to alter the signals.
Treatment of a series of chars with hydrogen iodide gas for sixty hours resulted in
reduction of the signal strength and broadening of the signals as shown in Table 1. The broadening
for the 300°, 400" and 500" chars was moderate but that of the exchange narrowed chars (600"
to 700") was extensive. Cold 0.1 M sodium thiosulfate produced no further change in the signals
but refluxing for twenty-four hours with thiosulfate restored the original signals14. The restored
signals of the 600", 625", 650° and 700' chars were sensitive to air whereas the restored signals
of the 300", 400" and 500" chars were not. The e.s.r. signals of the untreated chars refluxed
with 0.1 N Na, S, 0, were unchanged.

I
./
i
i
L \
Y'
J
J
/
51-5
TABLE I
Effect of Hydrogen iodide on Free Radical Content of Chars
Electron Spin Resonance Signals
spins/g in vacua and width in gauss* 1
Char
Hi Treotment Na, S, 0,O. 1 M
Temperature Untreated 60 hours Refluxed 24 hours
300
400
500
600
2.1 x 1010
4.9
2.3x 1019
6.1
5.1 1019
4.8
1.4 x 10"O
0.45
2.0~ 1017
7.9
5.3x 10l8
8.3
3.3x 1019
5.9
1.3 1019
19.1
8.3x 1017
5.76
2.bX 1019
6.3
5.4x 1019
4.1
1.4 x 1020
0.49
625 1.3 x 10"' 1.7 1019 9.8 1019
, 0.40 22.4 0.49
650 1.0 x 1020 4 . 1019 ~ ~ 1.0 x 1020
1.4 41 .O 1.2
700 7.4x 1019 1;2 1019 7.2 1019
1.4 43 2.0
* Width measured between points of maximum slope.
DISCUSSION
Two aspects of these results must be considered -the interaction with the halogens
which are typical electron occeptors and the interactions with hydrogen iodide. In agreement
with Wynne -Jones and co-workers, we are of the opinion that iodine, and to a lesser extent
bromine, can form donor-acceptor complexes with the chars and in so doing affect the e.s.r.
signals. Presumably the iodine and bromine act as acceptors and aromatic ring systems of the
char act as donors in these complexes. The low-temperature chars, 300" to 500°, having few
aromatic ring systems as large or larger than four or five rings1', are not affected by iodine or
bromine. The 600" char apparently has a high enough donor ability - low enough ionization
potential -to interact with iodine but not sufficient to react definitely witti bromine, which is a
much weaker acceptor. Both bromine and iodine readily form complexewith the 625", 650"
and 700" chars. The fact that the signals of the halogen affected chars are narrowed on cooling
to 79OK is also indicative of donor-acceptor interaction between halogens and aromatic moietiesle.
It is possible that lack of porosity is sufficient explanation of the failure of the 300" to 500" chars
to be affected by iodine',. However, the difference between iodine and bromine on the 600"
char suggests that the electrophilicity and not the size of the adsorbed molecule is important.

The line broadening at 79OK of the 300°, 400' and 500' chars treated with halogens
is not only indicative of no complex formation between halogen and the aromatic system but also
suggests that the initial signals of these chars are due to complexes between quinonoid acceptors
and carbocyclic donorsI6. This is further borne out by the effects of hydrogen iodide on these
In order to illustrate the relative importance of charge-transfer complexes in these
cham a plot has been made in Figure 3 of the number of free radicals vs. carbon content. There
has also been plotted the reciprocal of crystallite size in no./g. vs. carbon content calculated
from the work of Hirschl' and van Kre~elen~~ assuming that there is an average crystallite height
of three graphitic layers and that there is no unordered material. It should be noted that the
average height of the crystallites would be somewhat less than three in the low temperature cham
and possibly somewhat greater than three in the higher temperature chars. In any event, it is
seen that in the 600' to 650° range the number af crystallites clasely approximates the number of
free radicals, accounting for the results of Schuyera2. In fact these latter cham are apparently
composed of little but charge transfer complexes in close proximity to one another which explains
the intensive exchange narrowing of the e.s.r. signals of these materials. The decrease in free
radicals at higher temperatures is at least in part due to the decrease in number of cr)+staIlites
at higher charring temperatures.
Extendiw the conclusions on cellulase chars to coal and related materials, suggests
that e.s.r. signals in all the& materials result from donor-acceptor complexes which, while
reaching a maximum of importance in anthracites and 600' to 70O0 chars, may still be of
significance in lower rank coals and humic acids. Such a conclusion find, suppolt from a number
of other aspects of the behavior of coal and humic acids. Charge-transfer forces between coal
and solvent can be used to explain the order of effectiveness of coal solvents. Thus catechol
and anthracene oils, which am donor molecules, are very effective in extracting or dispersing
coal OT humic materials. Charge-tmnsfer equilibria can also rationalize the failure of coal
extmcts once dried to redissolve completely in the original solvent used for extraction. FIW
,
I
I

z
I
51.7
radicals in coals and chars are physically affected by oxygen, as indicated by line broadening of
the e.s.r. signal, but are stable to chemical attack. Such behavior could be explained as due
either to spin-spin coupling of the oxygen molecule with the free electron of the charge-transfer
complex or to formation of new donor-acceptor complexes between aromatic centres and oxygena4.
The so-called molecular weight of humic acids and cool extracts should be re-examined
in the light of the concept of donor-acceptor complexes. Thus molecular weight determinations
of humic acids in catechola5, a substance conceivably capable of breaking up dbnor-acceptor
complexes of humic acids, resulted in a low molecular weight estimate. Determinations on apparently
similar material in sulfalane, a poor complexing agent, produced very much higher estimates of
molecular weight 6.
finally, cools and related materials have a well-known broad absorption band at
1600 cm -l in the infrared region which has defied interpretation. Donor-acceptor phenomena
have been shown to produce strong and broad absorption in this regi~n"~~"~ in aromatic systems,
as shown in Figure 4, and hence may be the cause of this band in coal and chars.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
REFERENCES
Honda, H., and Ouchi, K., Rep. Resources Inst. Japan No. 1, 1952.
Adamsan, A.F., and Blayden, H.E.,Proc. of the 3rd Conference on Carbon (1957),
Pergammon Press, London, 1959, p. 147.
Berkowitz, N., Cavell, P.A., and Elafson, R.M., Fuel 40, - 279 (1961).
Ingram, D. J.E., and Bennett, J.E.P., Oil Mez. - 45, 545 (1954).
Uebersfeld, J., Etienne, A., and Combrisson, J.,Nature - 174, 614 (1954).
Winslow, F.H., Baker, W.O., and Yager, W.A.,J. Am. Chem. Soc. - 77, 4751
(1 955).
Akamatu, H., Mrozowski, S., and Wobschall, D.,Proc. of the 3rd Conference on
Carbon (1957), Pergammon Press, London, 1959, p. 135.
Singer, L.S., Spry, W. J., and Smith, W. H., Proc. of the 3rd Conference an Carbon
(1957), Pergammon Press, London, 1959, p. 121.
Ingram, D.J.E., Free Radicals as Studied by Electron Spin Resonance, Butterworths,
London, 1958, p. 213.
bbersfeld, J. and Erb, E., Proc. of the 3rd Conference on Carbon (1957), Pergammon
Press, London, 1959, p. 103.
Antonowicz, K., Proc. ,of the 5th Conference on Carbon (1961), Pergammon Press,
London, 1963, p. 61.
Mrozowski, S.,Proc. of the 4th Conference on.Carbon (1959), Pergammon Press,
London, 1960, p. 271.

13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
518
Campbell, D., Jackson, C., and Wynne-Jones, W.F.K.,Nature - 199, 1090 (1963).
Toland, W.G.,U.S. Patent 2,856,424 (1958).
Hirsch, P.B., Proc. Roy. SOC. (London), - A226, 143 (1954).
Base, M. , and Lobes, M.M., J. Am. Chem. SOC. - 83, 4505 (1961).
Mitchell, H.K. and William, R.J.,J. Am. Chem. SOC. - 60, 2723 (1938).
Deno, N.C., Freedman, N., Hodge, J.D., MacKay, F.P. and Saines, G., J. Am.
Chem. Soc. - 84, 4713 (1962).
Given, P.H. , and Peover, M.E., J. Chem. SOC. 394 (1960).
Voet, A.C. and Teter, A.C., Amer. Ink Maker 38, No. 9, 94 (1960).
Dulov, A.A., Slinkin, A.A., Rubinshtein, A.M., and Kotlyarevski, I.L.,lzvestia
Akod. Nauk S.S.S.R. Seriya Khim. No. 11, p. 1910 (1963).
Schuyer, J.,Brenn. Chemie. - 37, 74 (1956).
van Krevelen, D.W. and Schuyer, J., Coal Science, Elsevier Publishing Co. ,
London (1957), p. 176.
Tsubomura, H., and Mulliken, R.S.,J. Am. Chem. SOC. 82, 5966 (1960).
Polansky, T.S., and Kinney, C.R.,Fuel 1, 409 (1952).
Wood, J.C., Moschopedis, S.E., and Elofson, R.M.,Fuel - 40, 193 (1961).
Elofson, R.M., Anderson, D.H., Gutowsky, H.S., Sandin, R.B., and Schulz,
K. F., J. Am. Chem. SOC. - 85, 2622 (1963).
Sandin, R.B. , Elofson, R.M. , and Schulz, K.F., J. Org. Chem. - 30, 1819 (1965).
-

J 519
i
I
\
d
i
L
300 400 500 600 700
CHAR TEMPERATURE 'C.
IO#'
300 400 500 000 100
CHAR TEMPERATUR€ 'C
Effect of I, and Br, on line widths of
cellulose chars measured in vacuo at
room temperature.
Figure 2.
rn no treatment
A treated with 0.1 M I, in CCI,
treated with 0.1 M Br, in CCI,
Effect of , and Br, on spin concentrations
in cellulose chars measured in vacuo at
room tempera tu re
m no treatment
A treated with 0.1 M I, in CCI.
treated with 0.1 M Br, in CCI

W
L,
Z
6
m
(1:
0
6
d j o 2 ' 1
0.0
g 0.2
I I
0.4
I700
I
1400 1700
I
I400
FREQUENCY (CM-I)
6
Figure 3.
Comparison of the number of free radicals
with the size of crystallites in cellulose chars.
A.
8.
.number of free radicals per gram vs.
Carbon content
number of crystallites possible per gram
vs. Carbon content
Figure 4.
The influence of charge transfer forces
on infrared spectra.
A. Tetrakis [p-dimethylaminophenyl]
ethylene in fluarolube mull
'
B. Iodine complex of tetrakis
[p-dimethylaminophenyl 1 ethylene
in fluorolute mull
t.
I
i'
\

521
INFRARED SPECTRA OF OXYGEN-18 LABELED CHARS
R. A. Friedel
Pittsburgh Coal Research Center, Bureau of Mines,
U.S. Department of the Interior,Pittsburgh, Pa.
R. A. Durie and Y. Shewchyk
Commonwealth Scientific and Industrial Research Organization,
Chatswood, New South Wales, Australia
SUMMARY
Chars of four 0-18 labeled compounds were prepared. Infrared spectra were
examined for isotopic shifts. On the basis of a few published results on pure compounds
the shifts for C=O18 at 1600 cm-l were expected to be -20 cm-l or more. The
observed shifts are less than 5 cm-l; this result may indicate that chelated carbonyls
are not involved. However, very strong hydrogen bond'ing as in chelated
conjugated carbonyl structures may be responsible for the small shifts.
labeled chelate is being prepared at the CSIRO laboratories in order to investigate
this point.
- Chars
Introduction
!
An 0l8-
Since the early work at the Bureau of Mines on carbohydrate char&/ the
infrared spectra of chars ( 5 500" C) of various chemicals have been studied.m/
Spectral investigations of structure of chars have included studies of the effect
of elemental constitution; the effect of introducing oxygen into chars of hydrocarbons
indicated that spectral absorption bands, particularly the intense 1600 cm-1
band, were attributable to oxygenated structums.4'6II/
Isotope-labeling of chars
has been utilized; a successful study of deuterium labeling was carried out by comparing
infrared spectra of chars of an aliphatic hydrocarbon and a corresponding
deuterocarbon. From this study definite evidence was obtained for assignment of
the 730-910 cm-1 absorption bands to CH out-of-plane aromatic vibrations.=/
further indications were found that the most intense band in char spectra, 1600 cm-i,
was attributable to oxygenated structures.
A more direct investigation of the specific structures responsible for the 1600
cm'l band seemed desirable. Charring of compounds labeled with oxygen-18 was
judged to be a direct method with good chances for success. Three structures were
considered on the basis of their intense absorption, near 1600 an-1, to be possible
sources of the 1600 cm-1 band in chars:
(1) Carbonyl groups in ketones, quinones, or conju ated chelate structures
should show a shift to low frequency in going from GOLg to C=018; (2) aromatic
nuclei should produce no change in spectral frequency; (3) aromatic nuclei with
enhanced intensities due to oxygenated substituents on the ring should not undergo
any appreciable frequency shift in the 1600 cm-l region; a shift should occur in
the C-0 stretch region at 1100 cm-1.
Also

Coals
522
Studies of the 1600 cm-l band in char spectra may produce information on the
origin of the same band in the spectra of coals. Spectral data from chars of
carbohydrates would be particularly applicable for comparison with coal spectra as
carbohydrate chars produce spectra nearly identical to spectra of coals.u/ Unfortunately
018-labeled carbohydrates are not available.
Information on the intense 1600 cm-' band in coals has been obtained in studies
of absorption intensities for many pure compounds in an effort to determine what
structures could produce the 1600 cm-l absorption.j/ The principal type of struc-
ture found to have sufficient intensity is the conjugated chelated carbonyl group.
These groups can only produce the 1600 band if all or nearly all of the oxygen in
the coal is in the form of hydroxyl chelated carbonyl groups.
018-Labeled Compounds
The minimum spectral shift in the infrared spectra for a labeled carbonyl
group (C=018) relative to an unlabeled group (C=O16) is -40 cm-l, on the basis of
the simple ratio of the reduced masses. Hooke's law for the stretching vibration
of ~=016 is: u16 = 2rrc J =, where
p16 =
mC + m016 ; u = frequency, cm-1; k = constant;
%mol6
For C=OL8 the same equation applies; dividing,
'or ~ 1 = 8
Jl.05 '
For = 1600 cm-1, "18 = 1560 cm-li Au = -40 cm-l. If this 1600 cm-l band in a
char or coal spectrum is due to a simple carbonyl structure, the corresponding
labeled carbonyl group should produce an absorption band shifted to 1560 cm-l. If
the stretching vibration ,of the carbonyl structure is complex the simple value obtained
from the reduced mass calculation will differ somewhat from 40 crn-l. A
ketone investi ated by Karabatsos?/ showed an appreciable deviation; 2,3-dimethyl-
3-pentanone-018 shows a shift of 31 cm-l instead of the expected 40. This shift
is in the neighborhood predicted by FrancisEI for a mixed vibration composed of
about 80 percent C=O stretch and 20 percent C-C stretch. With conjugation the
shift does not change for the benzenoid carbonyl compounds:
benzoates, benzophenone, and benzoquinone.
Benzoic acid, two

523
AV, cm-1
Benzoic acid (monomer) -32- 111
111
Methylbenzoate -31-
Cinnamyl nitrobenzoate
Benzophenone
p -Benzoquinone
Benzoylchloride
-25- 11/
151
Benzamide -24-
The decreased shift for benzamide is attributable &greater participation of
other groups of the molecule in the carbonyl vibration.- Apparently the same
explanation may be given for the decreased shift observed for benzoylchloride. It
is surprising that the shifts observed indicate no particular increase due to con-
jugation with an aromatic ring.
(A large effect due to ring conjugation is observed
by comparison of the spectra of nitromethane-018 and nitrobenzene-018.16,17/)
Future investigations of more aliphatic ketones may produce different results.
The only published indication of the effect of conjugated chelation is on
benzoic acid dimer:
AU, cm-1
Benzoic acid dimer -20
Benzoic acid monomer -32
Thus hydrogen bonding is seen to produce a significant decrease in the OI8
shift of a conjugated carbonyl system. It was anticipated that the spectra of chars
having carbonyl groups enriched in 0l8 may shaw shifts for the 1600 cm-l band of
-20 to -30 cm-l. Such shifts in char spectra would.be measurable, even though absorption
bands are broad.
Experiment a 1
On the initiation of this work only three usable 0l8-labeled organic compounds
were commercially available. They were investigated in the following order:
Linoleic acid, phenol, and benzoic acid; sodium phenate and sodium benzoate-018
were also prepared and charred. First the unlabeled compounds were charred under
various conditions in order to determine the most favorable condition for produc-
ing intense 1600 cm-l bands.
Temperatures required to eliminate the absorption of
monosubstituted benzenes in order to produce coal-like absorption bands in the 730-
900 cm-1 aromatic band region were usually so high that other spectral details,
including the 1600 cm-I band, were weakened.
Samples of OI6 compounds were placed in a glass tube or a steel bomb, sealed
or unsealed, in vacuo or under nitrogen. Amounts of sample used were 50-100 mg
because of the scarcity of OI8 samples. Pyrolysis temperatures were varied from
350" C to 620" C; times were varied from 1 to 58 hours.

L in ole i c a c id - 0'
Two samples were investigated: (a) A sample of 22 percent 018-enriched acid
(kindly given by Dr. A. Miko, of the Yeda Research and Development Co., Rehovoth,
Israel), and (b) a commercial sample of 38 percent enriched acid. The best condi-
tions found for obtaining 1600 cm-l bands were 400° C, 2 hours, under 1 atmosphere
of nitrogen in a sealed pyrex tube. The 22 percent enriched sample was donated,
with the warning that polymerization probably had occurred; it is ais0 possible
that polymerization occurred in the 38 percent enriched sample. Yeda has taken
linoleic acid-018 off the market.
Phenol-018, 86 percent
Pyrolysis of phenol of 86 percent OI8 enrichment was carried out at tempera-
tures near 550" C for about 18 hours in a sealed tube under nitrogen.
Sodium phenate
The 0I8-labeled phenate was not prepared because of insufficient phenol-018.
Sodium phenate-016 was prepared from ordinary phenol and charred. Conditions were
similar to those used for phenol.
Benzoic acid-0I8, 95.6 percent
Conditions were similar to those used for phenol.
Sodium benzoate-0I8
A nitrogen atmosphere and a temperature of 500' C were used. Pyrolysis times
were about 24 hours.
In all samples except the sodium phenate and sodium benzoate, char was ob-
tained in the form of a thin film on the walls of the sealed tube. These films
were sufficiently thin to give spectra of good quality directly; data obtained from
such films was qualitative or semi-quantitative as thicknesses of the film varied
considerably. In all cases spectra were also obtained by means of KC1 pellets
which were better for quantitative absorption measurements.
Spectra were determined first on gases produced. Then the samples were washed
with hexane and/or benzene and the spectra of the soluble products were obtained.
The benzene-insoluble char was then investigated.
Ultimate analyses of all of the chars are given in table 1.
Results and ,Discussion of Results
Results obtained on the four substances investigated show that the spectral
shifts are very small or negligible (table 2); no shift was definitely detectable
for any of the chars. In figure 1 the spectra of chars from benzoic acid-0I6 and
-0l8 are given; they are practically identical, including the positions of the
two 1600 cm-l bands.
1600 cm-I bands were obtained in all ,chars but in no case were they intense
bands. This is a possible indication that,carbonyl groups may not have been in-
volved. For linoleic acid and benzoic acid it is nearly certain that no oxygen-
containing group could be an important contributor to the 1600 cm-l band, as the
oxygen contents of the chars were found to be negligible. Hydrocarbon structures
could have produced the weak bands found.

Y

526
Table 1.- Ultimate analyses of chars
c - . H ' " 0
" ' State
Table 2.- Negative spectral shifts of oxygen-18 labeled chars
Spectral shift,
cm-1 11
Linoleic acid-0I8

The phenol chars also showed no shift from lbOU cm-', even though one char contains
20 percent oxygen. It is not surprising if the pyrolysis of phenol at 550" C
does not produce carbonyl groups. The 1600 cm-l band observed in the spectrum of
the phenol char must be due to aromatic absorption enhanced by oxygen-containing
substituents on the aromatic rings (ethers and phenols).
The char of sodium benzoate also shows a negligible shift. If was the most
likely possibility for producing an isotope shift, as the elimination of oxygen
encountered in the chars of linoleic and benzoic acid does not occur for the char
of sodium benzoate. The sizeable percentage of oxygen present in the char, 4 per-
cent, is likely to exist in the char as carbonyl groups of some sort; even so, no
appreciable shift is observed.
The lack of appreciable shift in the spectra of 018-labeled chars may be attributable
to extremely strong association (chelation, or other). With the sizeable
oxygen content of some of these chars there are presumably some carbonyl groups pre-
sent.
Unfortunately no O18-labeled pure compounds having very strong hydrogen-
bonding have been investigated. As mentioned above, the spectrum of benzoic acid-
OI8 shows a smaller s ift for the associated dimer acid (-20 cm-l) than for the free
monomer (-32 cm-l).g7' These results are for a rather weakly associated acid dimer;
the vibration of the C=O group in the dimer obviously involves participation of
other groups in the molecule. For a more strongly hydrogen-bonded chelate, as
acetylacetone, the participation of other groups in the molecule would be greater
and the shift probably would be less. On this basis the small shifts in chars are
explainable. Certainly the negligible shift observed for chars does not exclude
the presence of C=O groups in structures producing the 1600 cm-1 band. In order to
investigate the effect of strong chelation on the C=OI8 shift preparation of dibeneo-
ylmethane, H5Cg-C018-CH2-C018-C6H5, is being carried out at the CSIRO laboratories.
Little or no e018 shift is expected. If an appreciable shift is observed for this
compound it w ill be necessary to conclude that chars do not contain strongly chelated
carbonyls.
Further studies will be carried out on labeled chars as other Ol8-containing
compounds become available. It is expected that prices of these compound will de-
crease greatly, for cheaper methods of preparation are being deve1oped.d
References cited
1. Friedel, R. A., and M. G. Pelipetz. Infrared Spectra of Coal and Carbohydrate
Chars. J. Opt. SOC. of Am., v. 43, No. 11, November 1953, pp. 1051-1052.
2. Friedel, R. A., and J. A. Queiser. Infrared Analysis of Bituminous Coals and
Other Carbonaceous Materials. Anal. Chem., v. 28, 1956, pp. 22-30.
3. Friedman, S., R. Raymond, L. Reggel, I. Wender, and R. A. Friedel. Chemistry
and Structure of Coal. Formation of Coallike Chars From Pure Organic Com-
pounds. Abstracts, Am. Chem. SOC. meeting, Dallas, Texas, April 1956.
4. Friedel, R. A. Spectra and the Constitution of Coal and Derivatives. Proc.
4th Carbon Conf., Univ. of Buffalo, Pergamon Press, London, 1960, pp. 320-336.
5. Brown, T. H., and Turkevich, J. A Study of the Low-Temperature Carbonization
of d-Glucose Using Electron Magnetic Resonance and Infrared Absorption.
Proc. 3rd Conf. on Carbon, Pergamon Press, London, 1959, pp. 113-120.

528
6. Friedel, R. A., and H. Retcofsky. 'Spectral Studies of Coal. Proc. 5th Biennial
Conf. on Carbon, The Pennsylvania State Univ., Pergamon Press, London, 1963,
V. 2, pp. 149-165.
7. Friedel, R. A. The Use of Infrared Spectra of Chars in Coal Structure Research.
Applied Optics, v. 2, No. 11, Nov. 1963, pp. 1109-1111.
8. Friedel, R. A., R. A. Durie, and Y. Shewchyk. The Origin of the 1600 Wavenurn- i
ber Absorption dad in the Infr

529
Recovery of Nitrogen Bases with
Weak Acids
by
T. F. Johnson, P. X. Masciantonio
and J. H. Schelllng
Applied Research Laboratory
United States Steel Corporation
Monroeville , Pennsylvanid
Introduction
The recovery of nitrogen bases is o'f particular interest. in
chemical plants that operate in conjunction with coal-carbonization
facilities.. . Traditionally, nitrogen bases' such as ammonia, pyridine, and
quinoline have been recovered by processes that involve acid extraction
followed by neutralization with caustic soda or lime. , 2, Such processes
have inherent disadvantages with regard to excessive consumption,of acids
and bases as well as concomitant waste-disposal problems'.
Recently, several processes have been developed for nitrogenbase
recovery by regenerative cyclic techniques. For example, recovery
of ammonia from coke-oven gas and pyridine from coke-oven light oil have
been reported.. * 5, These processes for recovery of nitrogen bases are
characterized by the use of an aqueous acid solution for the extraction
step followed by a high-temperature stripping, or springing, operation
that liberates the nitrogen bases and simultaneously regenerates the acid
extractant. The selection of an appropriate acid extractant for a
. '
particular nitrogen-base system is an important factor in designing or
developing a regenerative process. Previously reported studies3 , ')
have demonstrated that mqnoammonium acid phosphate is a suitable extractan,t
for ammonia recovery, whereas either phosphoric acid or monopyridinium
sulfate is adequate for pyridine recovery. However, the theory for
selection of appropriate reagents for regenerative nitrogen-base recovery
has not been developed and, consequently, the general applicability of
this recovery technique has not received significant attention. It is the
object of this paper to demonstrate that readily available thermodynamic
data can be used to determine the feasibility of a regenerative technique
for recovery of partkular nitrogen bases:'The discussion is based on
the behavior of particular acids during the extraction and springing
operations. Data are also presented on the extraction and springing steps
for recovery of pyridine and quinoline' from coal-tar fractions.
. .
Discussion. of the Regenerative Technique
The general scheme for a regeneratlve recovery process for
See References.
I

egeneration. The system can be seen in Figure 1 as a generalized flow
diagram. Usually, nitrogen bases are recovered from streams in which
they are present in low concentrations (about 0.1 to 5%), such as coke-oven
light-oil streams.
Selection. of Reagents for Regenerative Systems
The selection of an appropriate acidic extractant for a particular '
nitrogen base 1s the initial problem encountered in developing a regenerative'
process. Although such factors as solubility, volatility, decomposition, and
side reactions must be- considered, the acid component essentially must be
Free Hydronium Protonated Water
Base Ion Base
The temperature dependence of the equilibrium constant follows
the expression
where
AH = the heat of reaction, and
R = the gas constant.
I
I
,
\,
(2) I
It is desirable that the ratio (KT-/KT~) be less than unity for a \i
regenerative system (AH, the heat of reaction, must be negative).
I
In an effort to develop suitable parameters for the selection of .
reagents for regenerative systems, the ionization equilibria of acids and I
bases have been examined. I
In a discussion of acid-base systems, the following equations are 1
applicable. _ _
Acid extraction of a nitrogen base (B) from a hydrocarbon Stream'
involves the equilibrium reaction
(
B + H~O+ = BH+ + H ~ O
(3) '
which can be represented by an equilibr
um- constant
I
I
!
I
1
i

I
1
I
i
i
I
The ionization constant of the base involves the equll~brlum
B + H20 = BtI+ + 011-
expressed by an ionization constant
5 31
I
Kb = lBHfl [OH-] (6)
f,. I 13 I
,/
I
,'
By usina the lonlzati@n const

532
and can be expressed in terms of Equation 8 to give
Therefore
[Bl =I?)
[BH+]lI2
I
!
I
i
(16) ' I
Thus, for a given concentration [BH'] in the aqueous extract, the concentra- ~
tion [B] of base is high for bases of low base strength (Kb).
From Equations 12 and 17, the following generalizations can be
made concerning the selection of an acid and.a base for recovery of nitrogen
i
bases by a regenerative process. To get good extraction, maximize [BH+) by
using an acid with as large a value for Ka as is possible, consistent with
other requirements of the process. To get good regeneration, maximize [B]
(a base with a small Kb value should be involved).
i
Accordingly, suitabie systems appear to involve an acid extractant
with a large 1
Ka value and a nitrogen base with a small Kb value. Consistent;
with the above, the ApK was examined as a useful guide in selecting compo.-
1
nents 'for regenerative systems, . ,
where
pK = -log K (19) 1
ApK = (pKb -pKa) , and (20)
ApK is expressed as the absolute value of (PKb - pKa). It is
anticipated that the larger the ApK of a system, the better it w ill function!
in a regenerative recovery process. Therefore it appears that recovery of
nitrogen bases is facilitated by combinations of relatively weak bases and
relatively strong acids or relatively strong bases and weak acids. /
\
Systems for Recovery of Ammonia and Pyridine Bases
\
As mentioned above, several processes have previously been
reported for recovery of ammonia and pyridine bases from coke-oven product
streams. The data on ammonia and pyridine are consistent with the discussiol
presented above, Tables I and 11. Experimentally it has been demonstrated I
that, although ammonia can be recovered in a regenerative manner using
'q
monoammonium phosphate, neither phosphoric acid nor sulfuric acid can be
(
used (H2PO4- is suitable for NH3 recovery, whereas H3PO4, HS04-, and H2S04
are not applicable). '.
\
It has also been shown that pyridine can be recovered successfull
with either HjPOl or HSOq-; however, neither H2P04- nor H2S04 can be '6
employed.
i
1
k

1
i 533
1,
I
I
The use of the parameter ,pK does not appear to fit the observed
1 behavior of all the systems examined. However, the elementary ionlzation
ionization equilibrium theory explains certain inconsistencies with ApK
values. For example, in general, (1) systems in which both components have
\ pK values greater than about 6 are not useful since interaction is too weak
to effect extraction, (2) systems in which both componcnts,have pK values
less than about 6 are not useful since interaction is too strong tor
' dissociation of the acid-base salt during regeneration, and (3) when el ther
' component is a strong acid or a strong base, the system is not applicable
I because interaction is too strong for regeneration of the reagent.
I
I
/
Recovery of Pyridine and Quinoline from Coal-Tar Fractions
A regenerative system for recovery of pyridine from coal-tar
fractions has been examined on the basis of the above consideration ot ApK.
Values of ApK for the pyridine - HS04- and pyridine - H3P04 systems are
6.85 and 6.64, respectively, Table 11. Each system was examined to determine
I
the behavior of these acids in the extraction and regenerative steps. Data
on the quinoline regenerative recovery system are presented for comparison.
Ionization data on quinoline (Kb = 6.3 x pK = 9.87) suggest that
'
either HS04- or H3PO4 could be used as the sxtractant.
The respective values
of ApK are 7.95 and 3.22. Since HSO4- gives a larger value of ApK, the
recovery of quinoline with quinolinium acid sulfate solution as the extractant
was examined for comparison.
Experimental
Data on the extraction of pyridine and quinoline bases were
obtained by using conventional laboratory equipment consisting of a stirred
1-liter 3-neck flask equipped with a bottom outlet stopcock. Single-stage
extraction tests were conducted over the temperature range 0 to 55 C. The
data on the pyridine -,sulfuric acid system are presented in Table 111, and
data on the pyridine - phosphoric acid and the quinoline - sulfuric acid
systems appear in Tables IV and V, respectively. The extraction data
indicate that the nitrogen bases can be concentrated in the aqueous phase
by the extraction technique.
To examine the vapor-liquid equilibrium data for the pyridine -
phosphoric acid system, the pyridine - sulfuric acid system, and the
quinoline - sulfuric acid system, solutions were prepared of various
strengths. of acid and concentration of the respective nitrogen bases. Acid
strengths covered the range 20 to 40 percent and base-to-acid mole ratios
varied from 0.6 to 1.9.
An Othmer equillbrium still was used to determine concentrations
of pyridine and quinoline in the vapor and liquid phases for the various
sample solutions.
Discussion
The Othmer equilibrium data indicate that pyridine concentration

534
in the vapor increases with increasing concentration of acid (consistent wit
Equation 17) and, at a given level of acid strength, with increasing pyri-
dine-to-acid mole ratio, see Figures 2 and 3, Tables VI and VII.
Although it was expected that quinoline should behave similar to
pyridine, such was not the case. Quinoline was realized in the distillate i
of the Othmer still; however, the quinoline concentration lis significantly
lower than pyridine for a given mole ratio of base to acid. Two factors, ,~
low vapor pressure and low solubility, cause problems in the quinoline I
system. Because of the relatively poor solubility of quinoline in water ;
-(0.08 g/100 ml at.25 C), recovery of quinoline as a distillate product
depends on the limitations of a steam distillation system. Accordingly,
the distillate composition is a function of the vapor pressure of quinoline ':
at the boilihg point of the mixed liquid phase. Because of the low volatil-
ity of quinoline, only about 2 weight percent quinoline is realized in the
distillate during the regeneration step. Thus, typical Othmer equilibrium
data are not realized for quinoline.
Therefore, although quinoline can be recovered by a cyclic 1
regenerative process, the system is not analogous to the pyridine system.
The quinoline system appears suitable on the basis of ApK data, but other
factors such as solubility and volatility must be considered for practical
application.
Summary
A method for selecting reagents for recovery of weak bases from
hydrocarbon streams has been developed, based on thermodynamic ionization
equilibrium constants. An,acid can be selected for recovery of a given
base according to the value of ApK calculated for the system. Experimental
data have been presented on the pyridine - sulfuric acid system, the
pyridine - phosphoric acid system, and the quinoline - sulfuric acid system
and demonstrate the utility and limitations of the method.
The technique should also be applicable to development of systems \
When it is required to effectively remove nitrogen '
for recovery of acids from effluent streams by extraction with selected
nitrogen base reagents.
impurities in conjunction with a subsequent refining operation on a hydrocarbon
stream, a regenerative process may also be applicable; however,
provision must be made for adequate removal of nitrogen bases by using a
sufficient number of extraction stages.
References
1. H. H. Lowry, "Chemistry of Coal Utilization," Vol I1
John Wiley 6 Sons, Inc., New York, 1945.
2. P. J. Wilson and J. H. Wells, "Coal, Coke and Coal Chemicals,"
McGraw Hill, N. Y., 1950.
'I
I

t
'.
I
1 3.
\ 4.
I 5.
/
i
\
t
J
1,
535.
H. L. Stewart, U. S. Patent 2,410,906, November 12, 1946.
R. D. Rice, U. S. Patent 3,024,090, March 6, 1962.
G. D. Kharlampovich and R. I. Kudryashova, Coke and Chemistry (USSR),
- 2, 29, (1964). I

NH3
NH3
NH3
NH3
Pyridine
Pyridine
Pyridine
Pyridine
536
Table I
Data on Regenerative Recovery Systems
for Nitrogen Bases1) --
Nitrogen Bases Extractant
Ka-
6.23 x
7.52 10-33)
1.20 x 10-22)
-4)
6.23 x
7.52 x 10-
1.20 x 10-Z2)
33)
-4)
1) Handbook of Chemistry and Physics 44th Ed., Chemical Rubber
Publishing Co., Cleveland, Ohio, 963.
2) Ka for second hydronium-ion dissociation.
3) Ka for first hydronium-ion dissociation.
4) Completely dissociated.
Sys terns
NH3- (NH4) H2P04
NH3-H3P04
NH3-NH4HS04
NH J-H~SO~ Pyridine- (PyH) H2P04
Pyr idine-H 3P04
Pyridine- (PyH) HS04
Pyridine-H2S04
* ApK Not applicable.
' Table I1
ApK Values for Acid-Base Systems
Q!i
2.46
2.63
2.83
*
1.55
6.64
6.85
*
1
Kb-
1.79 10-5
1.79 x 10-5
1.79 10-5
1.79 10-5
1.71 10-9
1.71
1.71
1.71
Suitability of System
Yes
No
No
No
No
Yes
Yes
No

537
I Table 111
E
Extraction of Pyridine Bases from Lig$t Oil
i With Pyridinium Sulfate Solution
L I
I
I
i
I
1,
Test No.
1
2
3.
4
Mole Ratio of Pyridine in Pyridine in
Pyridine to H2E4 Washed Light Oil, % Extract, %
1.68
1.67
1.63
1.54
0.15
0.12
0.08
0.05
~ * In each test, light oil containing 0.44 percent of P-bases was
washed with a lean pyridinium sulfate solution (mole ratio 1.05)
prepared by adding crude P-bases to a 30 weight percent sulfuric
1 acid solution.
\
I
i
I
Table IV
Extraction of Pyridine Bases from Light Oil
(30 Percent Phosphoric Acid at 55 C)
I
Mole Ratio of Pyridine in Pyridine in Pyridine in
?est No. Pyridine to H3E4 Light Oil, % Washed Light Oil, %. Extract, %
I
I,) One-stage wash.
!) Two-stage wash.
I
.\
I'
i
i
I
t
3
1.0
1.0
0.70
0.92
0.95
5.67
2.10
4.60
4.60
4.60
0.44
0.40
0.24
0.59
0.58
29
29
28
27
19.5
19.2
14.5
18.1
18.6

Mole Ratio of
Test No. Quinoline to H2s4
1
2
538'
Table V
Extraction of Quinoline from a Tar Fraction
With Sulfuric Acid1 8 2,
I
1.46 '
1.41
Quinoline in
Washed Oil-, %
1.59
2.06
1) Tar fraction contained 12.1 weig t percent quino
2) Extractant contained 30 weight percent, H2S04.
Table VI
Quinoline in
Acid Extract, % '
me bases.
Pyridine Concentrations in Vapor at Equilibrium With
Refluxing H2E4 Solutions - Othmer Still Data
36.7
35.7
<
Initial H2SO4
Concentration, Mole Ratio of Pyridine Content
wt % Pyridine to H2E4 of Vapor, wt % \
20.4
19.9
30.9
30.4
30.6
40.3
1.63
1.95
1.60
1.81
1.98
1.60
4.7
21.1
10.4
23.8
41.6
21.3
i
I
!
\
r
'1

.. 539
Table VI11
Pyridine Conceqtrations in Vapor at Equilibrium With
I
Refluxing H3pO4 Solutions - Othmer Still Data
Initial H3P04 Pyridine Content
Concentration, Equivalent Ratio of of Vapor,
wt % Pyridine to Acid wt %
10.45
10.44
10.41
19.88
20.42
20.40
20.34
31.20
30.38
31.08
30.98
31.06
0.60
0.80
0.98.
0.40
a 0.60
0.81
0.97
0.60
0.80
0.98
0. 9ga)
0.99b’
a) Pyridine contained 23.4 percent picoline.
b) Pyridine replaced with picoline.
c) Analyzed as picoline.
0.54
2.08
6.86
0.0
0.87
4.72
13.84
1.11
6.52
21.46
19.92
21.65‘)

IXPROFJCTION
D; M. Mason
Institute of Gas Technology
Chicago, Illinois
Luminescence is the emission of noneguilibrium or nonthermal radiatio;
as opposed to incandescence. A number of materials have been observed
to l,mine?:ce !:.hen placed in a r'lame. This phenomenon, called candolurr.ineszence,
cqst be distirguished from thermoluminescence - in which
a xat?rial is excite? by radioactivity or other means but luminesces
~nl:: 7.h2n rai~~c to a higher temperature - and from incandescence and *
lurnirescxxe of the flame itself. The blue luminescence of an aerated
!i:Cthtln.- flax:. is an example of the latter. Candoluminescence is of
- i nt;::r?st from two standpoints - that of obtaining a better understandin-
cf flmes 2nd flme-solid interactions, and secondly the possibilit:;,
though remot?, of developing a new.method of gaslighting. The
wwl: reported here was supported in part by the American Gas Association.
Cm5nliuninescence has been known and studied for many years. It
vas probably first reported by Balmain in 1842 (3) in his description
of th? material boron nitride, which he was the first to prepare.
Apparently the phenomenon vas not recognized with other materials until
it 1uas reportej by Donau in 1913 (5). The subject was reviewed by .
L. T. Xinchin in 1938 (8) and by E.C.W. Smith in 1941 (11).
The materials that luminesce in flames share the characteristics
of phospiizrs generally; that is, they are comprised of a colorless
crystalline host or matrix material into which a small concentration
of foreign atoms or ions, called the activator, is incorporated. The
radiation is typically continuous and extends over only a small wave-
length range, thus having a definite color. The color is usually
characteristic of' the activating ion and can also be obtained with
2tkr means of excitstion, such as ultraviolet light or cathode rays.
?>:citation by a hydrogen flame has been used in most studies of
candolminescence, although Tiede and Buescher reported that, in ad- \
dition to hydrogen, flames of ethyl alcohol, hydrogen sulfide, and
carbon disulfide excited the luminescence of boron nitride (14). EXCI- 1
tation by the flame of city gas has also been reported (9).
The existence of the phenomenon at relatively low temperatures has
not been seriously questioned. The temperature range here is from
the temperature attained when a flame impinges on a thin film of phosnhor
spread on a cool metal support up to a bright red heat. This
upper limit is based on the observation of Tiede and Buescher (14)
that blue luminescence occurred when a flame touched boron nitride in
a carbon boat heated electrically to redness.
E. L. Nichols (10) claimed that at much higher temperatures he had
rsbserved light outputs from materials heated in an oxyhydrogen flame
that :.rere several times greater than the light output from a blackbody
at thc- same temperature. Apparently this work was inspired by the
,
\
1
I
L

limelight which was produced bG4keating a cylinder of lime with an
oxyhydrogen flame. The fact that a fresh portion of the lime had to
be brought to the flame from time to time was interpreted to mean
that somethm other than thema1 radiation was inmlved (8).
Neunhoeffer (3) reports that Lhe natural chalks from which the lime
was Prepared contain rare earths. If this were known to Nichols, it
would no doubt have suggested to him that the rare earths were activators
in a luminescence process. However, the work of later investigators
indicated that the limelight was similar to the light of the
Welsbach mantle in that both are only excited thermalky.
Nichols applied his "Dhosphors" on ceramic adjacent to a uraniumoxide-coated
8rea. and heated both areas evenly with an oxyhydrogen
flame. Observations were made of both areas with an optical pyrometer;
the uranium oxide vas taken to behave essentially as a blackbody and
to have the same temperature as the companion material. Many mixtures
Of colorless oxides uith small amounts of rare earths, or with some
other elements that were also regarded as activators, had a greater
emittance than uranium oxide. E.C.W. Smith (11) repeated some of
Nichols's experiments and obtained similar emittance readings. He
then proceeded to determine or approximate the actual temperature of
the two coatings - a matter of some difficulty. By means of thermocouples
made of extremely fine wires, he was able to show lwge differences
in the temperatures of the two coatings - differences suffi- *
cient to conclude that at incandescence temperatures there was no
emission other than thermal. This conclusion has been confirmed by
Sokolov and his eo-workers (12), (13).
In hydrogen flames, at least, recombination of hydrogen atoms is I
generally accepted as the major source of energy for the low-temperature
excitation (1) , (7); earlier Donau (5) and Nichols (10 considered
an oxidation-reduction mechanism, and Neunhoef fer (9 1 thought
that electrons in the flame were responsible. Smith, and Arthur and
Townend (l), (2) investigated this question with phosphors composed
of calci1.m oxide activated with manganese, antimony, or bismuth. They
found that:
1. Activated oxides that luminesce in hydrogen flames also luminesce
when subjected to the action of hydrogen atoms produced in an electrical
discharge. When subjected to the latter action and heated
electrically, the color of the luminescence approached the color
produced in the flame. Presumably, the temperature of the heated
oxides approached the temperature of oxides in the flame.
2. The intensity of hydrogen flame luminescence increases when the
hydrogen is burned under reduced pressure. This effect was presumed
to be caused by a decrease in the number of three-body
collisions. These collisions remove hydrogen atoms by recombination.
3. The luminescence is strongest in hydrogen, erratic and very faint
in town gas, and absent in carbon monoxide flames. It is also
absent in methane and ethylene flames tested at pressures from
atmospheric to 10 em Hg.
4. The luminescence is extinguished when small amounts of hydrocarbon
gases or vapors are added to the hydrogen.

-4:c~rciiny to recent studies o ;i42 flame mechanisms, hydrogen atoms are
m>zs?Iit i_n the methane-air flame as well as in the hydrogen flame (15).
This it aypna!-eC! reasonable to search for phosphors that candoluminesce
in th3 :?,t!isnL7-air flame. This search and the study of the characteristl-s
of this phmonenon were the objectives of our investigation.
?.!-I; c pial s
R-rc zarthc: vere obtaineci as ?..?.'?:I pure oxides from Lindsay i
Ci?c.:-:iccl Pivision of Amrican lotash and Chemical Corporation. Other
zhmi. :?-ls vert rcsearch grade.
Ca1ci.w OxiSe Phosphors
I Calcium and rare .earth nitrate solutions were prepared and mixed
in the required amounts. The mixed nitrate solution was slowly poured
intc hot mionium carbonate solution with stirring. The precipitate
vas Cilt:?md, mshed vitli hot ammonium carbonate solution, dried at
13C)°C, and ignited at 900°C for 30 minutes. -The pure rare earth oxides
:wr? prcpared by precipitation from the nitrate in the same manner. !
Yttrium Europium Tunastate - (YO. ~Euo. I)PO~WO~ - and Gadolinium
Europ.i.um Molybdate - ( Gdo. &uo. 1 >Os. Mooa
T!ie required amounts of the oxides (tungstic acid in the case of
tjqsten) wme mixed, heated in a platinum crucible at 1000°C for 2
hours, :-round with mortar and pestle, and reheated at 1000°C for 2
hours. The X-ray patterns of the two-products showed that in both a
new phase had been .formed.
Screcninp; Tests
I
Mounting strips about 2 cm wide and 10 em long were made of insulating
fire brick and of copper sheet. Phosphor powder, thick enough to hide '
the surface of the strip, was pressed on to it with a spatula. The '
nearl:i vertical face of the strip was observed as it was passed through
a methane-air flame (bunsen burner) and a hydrogen diffusion flame in
a dark room. (
Spectra
Thrne different methods of mounting the phosphors were used in
obtainiw the spectra. When we wished to observe the effect of temperature,
the phosphor was applied as a paste to a silicon carbide rod.
Pastes made with monomethyl ester of ethylene glycol seemed to adhere
after drying somewhat better than those made with other liquids. The
silicon carbide rod, about 6 mm in diameter, was held between two
vater-cooled electrical terminals to allow the electrical current
passing between the terminals to heat the silicon carbide rod and the
phosphor.
iYhen hydrogen was used as a fuel, the phosphor could be coated dry
on the fritted disk of a Gas-dispersion tube. The fritted disk was
operated as a porous-plat€ burner (Figure 1) without p-imary air.
Knoiin ai-,o~ints of other gases could be added to the hydrogen.
I
I
\
I
,(
I
1
I
b

I
I
1
543 - ,
The phosphor ijas coated on a water-cooled eo r plate when a
methane-air flame was to be used. The plate was 1 in. square x l/$ in.
thick wit!i trro l/k-in. copper tubes soldered to the back for cooling
water. Plating the face with silver was found to be necessary to prevent
slow Poisoiling of the phosphor by the copper. The phosphor vas
settled onto the plate from a water suspension. The plate was mounted
facing upwards at about 45' to the horizontal and the flame of a
1 National Welding Equipment Go. Type-3A blowpipe was d3,rected downward
jonto it.
I A Beckmail DK-2 spectrophotometer with an IP 28 photomultiplier was
used to record spectra. The backplate of the lamp housing was removed
to al1o:s direct entrance of the radiation from the excited phosphor
into the spectrophotometer.
A slit opening of 0.4 mm without an aux-
iliary light-gathering system gave sufficient enera. Varying light
intensity Prom the flickering of the flame was troublesome and made it
necessary to use the highest time constant and the slowest recording
rete of the instrument. The porous-plate burner gave the steadiest
radiation.
RZSULTS
Screening tests to discover phosphors that candoluminesce in methaneair
flames or in hydrogen flames were run on a number of commercial
phosphors and on others prepared in our laboratories. The latter were
principally rare earths - pure and in calcium oxide. The phosphors
were spread on fire-brick slabs and on copper strips - the latter to
hold down the temperature of the phosphor at least momentarily - and ,
tested by impi,agement in both hydrogen and methane-air flames. Sev-
, eral phosphors that candolumenesce in both flames irere found. Detailed
results are shown in Tables 1 and 2. These results should be regarded
as tentatlve rather than conclusive.
A few additional phosphors not listed in the tables were tested.
' Three General Electric silver-activated zinc sulfides (NOS. 118-2-3,
118-2-11, and 118-3-1) showed emission in the blue. Their spectra,
abtained with a hydrogen diffusion flame on a fritted glass disk,
showed that the emission was the band spectrum of S2 and that no cando-
)luminescence was present (6). The S2 was undoubtedly formed by decom-
position of the phosphor.
Zinc sulfide activated by silver and cop er
showed both candoluminescence and the band spectrum of S2 (Figure 27.
A boron nitride sample from Carborundum' Co., Electronics Division
showed weak, green luminescence in hydrogen and methane flames. Two
recently developed rare earth phosphors, (YO. s EUO. 1)203-W03 and
( Gdo. eEuo. 2) 203- MOOS were prepared and tested (4) . Both luminesced red
in hydrogen and metharle flames on the fire-brick strip.
,
Light emission in some of the screening tests may have had a source
other than candoluminescence. The emission from magnesium germanate
,was probably thermal. This is indicated by the color of the emission
(vhite to yellowish white), by the absence of a maximum of emission
intensity vith increasing temperature of the phosphor, and by equal
mission \?hen the phosphor is heated to the same temperature with or
eithout the flame. For this test, the phosphor was mounted on the
electrically neated silicon carbide rod, and temperature equivalence
'was indicated by the emission of the phosphor at 2.25 microns. Light
J
!
I

544
Table 1. LIGHT EMISSION OF Y"TRIUM OXIDE AND RARE EARTHS (Pure and
in CaO) IN HYDROGEN AND METHANE-AIR FLAMES
Element
Yttrium
hnthanm.
Neodymium
Samarium
GadOlinlum
Dyspmsium
Holmium
Erbium
Ytterbium
- Fure oxide.
-
Atomic
No. a
39 Y
57 la
59 Pr
60 Nd
62 9m
64 cd
66 Dy
67 H@
68 Er
70 Yb
Metal
Conc
in
CaO.
vtF
0.2
1.0
Po'
0.2
1.0
PO
. 0.2'
1.0
PO
0.2
1.0
Po
0.2
1.0
PO
0.2
1.0
PO.
0.2
1.0
PO
0.2
1.0
Po
0.2
1.0
Po
0.2
1.0
PO
ht msslon
3ith €L LU W-
On
cu
Trace
Trace
Weak
Trace
Trace
Weak
Trace
Trace
Nil
Trace
Trace
Nil
Trace
Trace
Nil
Trace
Trace
Strong
Trace
Trace
Nil
Trace
Trace
Nil
Trace
Trace
Nil
Trace
Trace
Nil
On
Fire
?M&
Weak
Trace
Trace
Weak
Weak
Trace
Stmng
Strong
Trace
Trace
Weak
Trace
Trace
Trace
Nil
Trace
Weak
Strong
Trace
Weak
Nil
Trace
Weak
Trace
Trace
Weak
Trace
Trace
Trace
Nil
on
cu
Nil
Nil
Trace
Nil
Nil
Nil
Nil
Trace
Nil .
Nil
Nil
Nil
Trace
Nil
Nil
Nil
Nil
Trace
Nil
Nil
Nil
Nil
Nil
Nil
Nil
Nil
Nll
Nil
Nil
Nil
- r
on
Fire
m
Nil
3:
Nil
Nil
Nil
Strong
strong
Nil
Nil
Nil
NiI
Trace
Trace
Nil
Nil
Nil
Weak
Nil
Nil
Nil
Nil
Nil
Trace
Nil
Nil
Nil
Nil
Trace
Nil
Color
Violet !
I
Violet
Green
and
Orange
Lavender 4
lavender
Green
-and
Orange I
Red
Red I
Red
Violet
Lavender
Red
Oee
orange
--
Violet
Violet
Red
Lavender
Blue
--
Aqua to Blue
Aqua
Red
Blue
White
Red
Lavender
Red
--
I
!
I
i
I
1

\
'i
2
w
8
?N4"?? I? IC9
t-OO+r(N 0 A N
545

emission in a few cases may be by decomposition of the phosphor,
es vas shom in the case of the silver-activated zinc sulfides.
The manganese-activated zinc phosphate (Sylvania No. 151) phosphor
appeared to give the most intense luminescence emission in methane-air
flames of my tested, and, accordingly, its light emission was investi-
-
aated further. \Then the phosphor was coated on a fritted disk and the
Eisk used as e porous-plate burner, a steady emission was obtained with
a hyerocen dif'fusion f'lame. However, no emission could be obtained in
this manner vhm the fritted-disk burner was fed with Qure methane or
vith a --thane-air mixture. Only when the phosphor was brought into a
i3dnsnn flair.? -;:as the light emission observed. We do not know the cause
sf this effsct. A greater concentration of hydrogen atoms in the hydro-
play a. role. The high diffusivity and other properties
hz't zause its flame to burn closer to the solid surface
rL or greate? importance.
Sssctre recor2ed with the Beclanan DK-2 spectrophoto.meter were used
to' zonpare the light emission of this phosphor in the hydrogen flame
vith its emission in the methane-air flame. The phosphor was coated
on 2 silicon carbide rod for this experiment. No significant differ- ,
ence other than that of the emission from the flame itself was observed
('izurc, 3 ) . I
T!ie effect of temperature on the light emission of the phosphor was
i nvasti.?at?5 by electrically heating the silicon carbide rod. With a
nethanerair flame impinging .on the phosphor-coated rod, and with the
spectrophotometer focused on the rod near one of its water-cooled ends,
the Li&t emission increased to a maximum with an increase in electrical .
hGatin5, then decreased to near extinction with further heating. This
agrees with the observation of Neunhoeffer (9) that the hydrogen flame.
candoluminescence of calcium oxide impregnated with various activators '
eshibit s temperature maxima.
When'the phosphor was heated to a still higher temperature, thermal
emission appeared. When the emission of the phosphor was corrected
'i
for the emission of the flame, no essential difference in emission at
high temperatures with and without the flame was evident. These experi- .'
ments indicated that the low-temperature light emission experienced
with this. phosphor is candoluminescence, but that candoluminescence is
not involved in the high-temperature emission.
I
This phosphor was'the only one of several tested that luminesced in
E carbon monoxide flame. The question of excitation in this flame by
high-energ;; species other than hydrogen atoms was not pursued.
Spectra of the luminesccnces from several phosphors excited by
hydrogen burning on the surface of a sintered glass disk were obtained, '
a& the effect of the addition of small amounts of carbon monoxide and
methan? to the fuel was observed. The.phosphors were calcium silicate '
aztivatec with lead and manganese (Sylvania No. 290), calcium oxide ,
activated vith ytterbium, calcium oxide activated with praseodymium,
and salcim oxide activated with samarium. The spectra obtained from
the f5rst three of these phosphors when they are excited with pure
hj-d?Ggen burning on a fritted disk are shown in Figure 4. The Sylvania 1,
KG. 24.9 phosphor emits with a broad peak in the green. Praseodymiumecti:retc.c
cal2im oxide shows two peaks - one at 595 millimicrons /
(p110:.1) and a weaker one at 490 millimicrons (blue-green). Ytterbium- 1
i

I
i
t
\
I/'
t
I
activated calcium oxide also sh&J two peaks - one at 375 millimicrons,
which is in the ultraviolet. Samarium-activated calcium oxide also
emits with a similar peak in the ultraviolet and another at 570 ,
mllllmicrons (yellow), as shown in Figure 5.
The effect of the addition of small amounts (1 to 10 volume percent)
Of carbon monoxide or methane to the hydrogen burning on the fritted
disk was different for each of these phosphors. Luminescence of the
Sylvania No. 290 phosphor was substanitally unchanged with addition of
as much as 105 of the carbon monoxide or methane. Lwipescence of the
Praseodymium-activated calcium oxide was not greatly affected by addition
of 1" of the carbon monoxide or methane, but decreased rapidly
with greater amounts.
obtained with 62 carbon monoxide or 846 methane. The luminescence of
the ytterbium-activated calcium oxide was even more strongly quenched
by the addition of carbon monoxide or methane. With these phosphors,
only the intensity of the luminescence was affected.
With samarium-activated calcium oxide, the spectral distribution of
the luminescence was also affected, as shown in Figure 5. Upon the
addition of carbon monoxide, emission by this phosphor in both wavelength
regions decreased in intensity, and, at about 3% carbon monoxide,
emission in the 375-millimicron region had virtually disappeared. The
570-millimicron peak, however, was replaced by two narrower emission
bands with peaks at about 560 and 595 millimicrons. With further addition
of carbon monoxide the intensity of this emission increased to a
maximum several times more intense than the original peak emission,
then decreased. These effects were not observed upon the addition of
methane to the flame. The variation of the intensity of the luminescence
Fn the two cases is shown in Figure 6.
At a late stage of the work, phosphors were deposited by settling
from a water suspension onto a water-cooled copper plate. A few spectra
were obtained in this way with methane-air excitation.
Ehhancement of the luminescence from the manganese-activated zinc
silicate (Sylvania No. 161) by the cooling of the copper plate was very
pronounced. (Figure 7 ) .
ACKNOWLEDGI4ENT
Only about 10% of the original intensity was
The author wishes to thank the American Gas Association for their
support and for permission to publish these results. John Hasenberg
did most of the experimental work. Peter Mrdjin identified the S2
bands in the spectra of the zinc sulfide phosphors.
. .
. .

LITERATUR2 CITED
i.
C.
-.
.' *
il.
5.
'2 .
7.
8.
9.
10.
11.
12.
13.
14.
543
Arthr, J. R. and Townend, D.T.A., "Fundamental Aspects 0;
Sombustion - Some Reactions of Atomic Hydrogen in Flames,
J. Inst. FLel 2-0, 44-51, 61 (1953) July.
t rtnur, J., R. and Ttwnend, D.T.A., "Some Reactions of Atomic
Hydrogen in Flames, in U.S. Natl. Bur. Standards, Circ. 523,
"Znergy Transfer in Hot Gases, ' I 99-110. Washington, D.C. :
.;ovt. Print. Office, 1954. I \
Bslrrain, W. H., ,"Observations on the Formation of CEmpounds of
Boron ana Sllicon With Nitrogen and Certain Metals, Phil. Mag. 2l,
2-3-7, (1342).
Borchardt, H. J., "Rare Earth Phosphors, J. Chem. Phys. 2, 1251
(-333) March 1.
Pcnau, J., "New Kind of Luminescence of Alkaline Earth Preparation,
Especially Calcium Preparations Which Contain Bismuth or Manganese
Brought Out by the Hy$rogen Flame; as Well as the Detection of .
Traces of the Latter, Monatsh. 2, 949-55 (1913).
Gaydon, A. G., The Spectroscopy of Flames, 244. New York: John
liiley, 1957.
Gorban, A. N. and Sokolov, V. A., "On the Question of the Physico-
Chemical Nature of Candoluminescence,'" Opt. Spectr. (U. S. S. R. )
Eixlish Trans;. 1, 1C:-65 (1959) August.
Minchin, L. T., "Luminescence of Oxides Under Flame Excitation, I'
Trans. Farada SOC. 35, 11163-70 (1939) ; "Luminescence of Substances
'Under Flame Gcitation, Trans. Faraday SOC. 2, 505-06 (1940).
Neunhoeffer, O., "The Hydrogen Flame as a Source of Electrons, 'I
Z. Physik. Chem. x, 179-85 (1951).
Nichols, E. L., Howes, H. L. and Wilber, D. T., "CF)thodo-lumines-
cence and the Luminescence of Incandescent Solids, Carnepie Inst.
Pubi. No. 348 (1928).
Smith, E. C. W. . "The Radiation From Solid Substances Under Flame
Sokolov. V. A., "Candoluminescence From the Viewpoint of the
Vavilox-Wiedemann Criterion and of Contemporary Physicochemical
Ideas, Izv. Akad. Nauk. U.S.S.R., Ser. Fiz. - 26, 514-17 (1962).
Sokolov, V. A., Grozina, I. S. and $orban, A. N., "The Nature of
Candoluminescence of Calcium Oxide, Izv. Tomskogo Polite-. Inst.
42, 253-56 (1958).
Tiede, E. and Buescher, F., "Inorganic Luminescent Phenomena. 11.
Luminous Boron Ni5ride (Balmain's Aethogen and the Flame Excitation 1
of Luminescence), & a, 2206-14 (1920).
<
'
i
I

I
15. ',Jise, H. and Rosser, W. ,,A., "Homogeneous and Heterogeneous Heactions
of Fh~e Intermediates, in Ninth Symposium (International ) On
Cmbustion, 733-42. New York: Academic Press, 19b3.

co, CH,,OR 1
-
OTHER ADDITIVE
FRITTED
1 GLASS
+ DISK
\LIGHT BEAM
\ SPECTROPHOTOMETER
L-KE-1
' PHOSPHOR
BURNER
DETAIL 1-4121
:-'L :.:?? 1. SCHEATIC REPRESENTATION OF THE, POEicJUS BURNER
CPPAR?TUS AWR LIGHT ENISSIOIi STUDIES
?:pire 2. Ei41SSION SPECTRUM OF ZnS: Cu:Ag EXCITED
' EY A HYDROGEN DIFFUSION FLAME

B 549
15- '-Jise, H. and Rosser, W. llA., "Homogeneous and Heterogeneous Reactions
Of Flame Intermediates, in Ninth Symposium (International) on
Canbustion, 733-42. New York: Academic Press, 1963.
I

FRIT 'TED
CO, CH,.OR 4
___
OTHER ADDITIVE
i
@)
FUEL
\ PHOSPHOR
BURNER
DETAIL A-b?Zl
Z..;.:T? 1. SCHIGiATIC REPRESENTATION OF THE PGROUS BURNER
."-PPAR3.TUS FOR LIGW E31ISSIGTJ STUDIES
?:;Lire 7. Ei4ISSION SPECTRUM OF ZnS: Cu:Ag EXCITED
' BY 4. HYDROGEN DIFFUSION F W

I
0
I I I I 1 I I I I 0 R
2gDOc- 0 8 8 : . F l O 0 N - 0 0
(3SNOdS3U 1133010Hd ljOj 03133kJU03Nn)
NOllVlClVU 03111W3 30 AlISN31NI 3A11Wl3U
8
P
0
al IC)
0
a R

340 350 300 400 4 50 500 600 700 800 900
WAVELENGTH, MILLIMICRONS 1 e.’
Figure 4. EMISSION SPECTRA OF PHOSPHORS EXCITED BY A
HYDROGEN DIFFUSION FLAME
\
300 400 450
A -A
WAVELENGTH, MILLIMICRONS * .e
Figure 5. EFFWT OF CARBON MONOXIDE ON THE EMISSION
SPECTRUM OF SAMARIUM-ACTIVATEE CALCIUM OXIDE

1
2.2
2.0
0.8
0.6
0.4
553
0 2 4 . 6 8 IO
CO OR CH, CONTENT, VOLUME K A-4771
Figure 6. EFFECT OF FUEL COMPOSITION ON THE CANDOLUMINESCENCE
OF SAMARIUM-ACTIVATED CALCIUM OXIDE
. .

0
554
WITH COOLING
(PLATE AT 40°C)
COOLING
4500 5000 6000
WAVELENGTH, A A-4973
Figure 7. THE EFFECT OF TEMPFBATUFE ON THE E3:ISSION SPECTRUM
OF SYLVANIA NO. 161 PHOSPHOR EXCITED BY
A METHANE-AIR FLcuvlE
i
I
1
1
‘I
I
\
I 1
1
\

555
A RAPID, SIMBLE METhOD FOR THE DETERNINATION
OF THE THERMAL CONDUCTIVITY OF SOLIDS
Neal D. Smith
Fay Fun
Robert M. Visokey
UNITED STATES STEEL CORPORATION
Research Center I
I' Monroeville, Pennsylvania
I
I
I
The rational design of equipment such as shaft coolers, heaters,
I and rotary kilns for the heating and cooling of solids requires that
the thermal properties of the solids be known. Thermal conductivity is
one of these properties that to measure necessitates elaborate equipment
I and tiqe-consuming techniques.
A rapid, simple method has been developed €or determining the
thermal conductivity of solids.' The solids can be either porous or non-
porous and of either high or low conductivity. If high-conductivity
, materials are tested, then both the thermal conductivity and heat capa-
I city caq be simultaneously measured by the method.
The procedure involves preparing a cylindrical briquette of the
.test solid th'at has a thermocouple located in the center. This bril
quette is heated to a constant temperature after which it is suspended
' in an open-end glass tube and cooled by a known flow of nitrogen or any
other nonreactive gas. The thermal conductivity is then computed from
(.a digital computer comparison of the cooling curves for the test solid
' versus a reference solid of known thermal properties and similar size
1 that has undergone the same heating and cooling cycle. The method was
validated by using the known thermal properties of lead, aluminum, and
silver and computing the theoretical cooling curves. The theoretical
, curves were in close agreement with the experimentally measured cooling
curves for these materials.
' marized as follows.
Theory
I The mathematical.basis for determining thermal conductivity by
the described method is discussed in a paper by Newman') and is sum-
Consider a cylindrical briquette as shown in Figure
( 1. The differential equation for unsteady state heat transfer by conduction
in the x-direction is (see nomenclature for definition of the vari-
1 .
1 For a briquette of thickness Za, the central plane being at x = 0 and
) assuming:
i
' 1) uniform temperature at the start of cooling of the initially hot
1 briquette

5 56
then t = to when e = 0
2) the final temperature of the briquette will be the temperature of
the surroundings : I
therefore t = t, when e = *go
3) there is no heat flow across the central plane because of symmetry: {
I
consequently -k (2;) - = 0 at x 0
The heat balance on the briquette surface is made by equating heattrans-
ferred to the surface by conduction with heat transferred from the surfac
by convect&,on.. In differential form, the heat balance is: '
-k dt = h (t - t,) at x = +a
. (a4 %
Newman') showed that the solution to Equations (1) through
in terms of a dimensionless temperature ratio Yx is:
ma
(I+$ ma2 +ma) cospn
where An= and
pn are defined as the first, second, third, etc., roots of
cendental equation:
Pn TAN pn - I /ma = 0
The surface to solid thermal resistahce ratio, ma,
ma = k/ha
and Xa is defined as: xq = ae/a2
where the thermal diffusivity is: a = k/p Cp
(5)
(5) expressed
the trans-
Similarily , considering radial heat transfer , the .radial briquettl
heat balance is
The initial condition equation is:
toto WHEN' 8=0
The final temperature equation is:
t= 1, WHEN 8. -
The boundary condition equations are:
-k (g)=O AT r=O
(7)
(12) i

557
and -k (%)= h (t-ts) AT f =R
Solving Equations (11) through (15) gives:
and Pn ar,e the first, second, third, etc., roots of the equation:
and . .
PnJI(fin) - I/mr Jo(Pn) =.O
The surface to solid thermal resistance ratio, mr is
I x, = a8/R2
~ and
is :
(18)
frIr = k/hR (19 1
The complete differential equation for the case shown in Fig. 1
the solution to Equation (21) is:
, eqn. 22 becomes: . .
I
If the center temperature defined at r = 0, x = 0 is tc, then
tC’+S
Yc= - = Yr
to ’tS
Y,
where y,
1
and YX are evaluated at r = 0 and x = 0.
i The preceding mathematical analysis shows that the rate of cooling,
lor change in center temperature for a cyclindrical briquette is a function
I of, time (8 ) , density ( e) , thermal conductivity (k) , the surface heat
) transfer coefficient (h) , specific heat (C,) and the briquette dimensions
I as expressed by Equation (23).
The experimental technique can now be described in terms of th;
~p~vious discussion. If the change in center temperature with time is Rsa-
,sued experimentally for a material of known thermal and physical prop-
‘erties (standard briquette) , the surface heat transfer coefficient can be
calculated from Equation (23), since it is the only unknown.
(20)

5 58
The surface heat transfer coefficient (h) is a function of the
flow rate of the cooling gas and the geometry and size of the briquette.
It is independent of all other physical, thermal, or chemical properties
of the briquette. Therefore, any other briquette having similar dimensions
and cooled at the same flow rate will have the same value for (h).
Once (h) has been determined using the standard briquette, the
thermal conductivity of any test material can be, determiped from Equa-
. tion (23) since all other variables are known.
A computor program has been written which through an iterative
process determines the best value of (h) which makesthe calculated
values for the dimensionless temperature ratio equal to the experimental
values obtained when the standard briquette is cooled.
With (h) deter&ned, another, computor program is run for the test
specimen. Thermal conductivity is now the unknown variable and through
another iterative scheme, the best value for (k) that makes the calcu-
lated and experimental values for the temperature ratios equal is found.
The input data for both programs consist of density, specific
heat, time, briquette dimensions, and several experimental values for
the temperature ratio. The output from the first: program (standard) is
the best value for (h). Using this value for (h), the second program
used to determine the k value €or any test material. If a highly conduc-
tive material is tested, then it is possible to determine its heat capa-
city since the solid thermal resistance will be small compared to the
surface thermal resistance. A transient heat balance can be written for
. the test solid cooling in a stream.of coolant gas.
VpCp dt = hA (t - ts)
de
In the above equation, t = tc since the thermal gradient in the solid is
'neglected. Integrating Equation (24) and using the dimensionless temperature
ratio, Yc gives:
Yc = exp(hA/pCpV) e (25)
Thus, if the internal solid thermal resistance is neglible, a plot of
the experimental Yc versus e data on semilog paper should be linear as
shown by Equation (25). The heat capacity, Cp, can be calculated from
the slope of the line for Yc versus e since (h) is the same as for the
standard briquette and the density, p, and total surface area, A, for
the test material are also known.
Materials and Experimental Work
A primary advantage of the transient technique for determining
thermal conductivities is the ease and swiftness with which the experi-
ment can be conducted.
In so far assunple preparation is concerned, any solid that Can
be briquetted, cast, or fabricated around a centrally located rigid
c

559
thermocouple (x = 0; r = 0) may ue tested. Finished test sazple ciriin-
ders should be approximately one inch in diameter, and one-haif inch ir.
height; however, other dimensions can be used.
Experimental Apparatus
The experimental apparatus (see Figure 2) consists simply of a
3-inch diameter glass tube approximately 3 feet in length. One end of
, the tube .is completely stoppered except for a one-half iAch circular
opening through which the coolant gas flows. The other end of the tube
I is open to the atmosphere. A small electric furnace is Lised to heat
1 the briquette, and an automatic single point temperature recorder con-
nected to the embedded thermocouple is used to measure the center temp-
erature of the briquette.
Experimental Procedure
The experimental procedure is the same for both the standard and
test briquettes. Either the standard (aluminum was chosen since its
thermal properties are well established), or the test briquette is con-
nected to the temperature recorder by way of the thermocouple lea&.
The briquette is heated until the center temperature has reached a con-
stant , predetermined value. The briquette is then quickly removed from
1 the furnace and suspended in the cooling tube with the cooling gas flowing
at a constant rate. The briquette is usually cooled to the temperature
of the cooling gas within 20 minutes.
Data Processing
I
i For the standard briquette , the experimental dimensionless temp-
erature ratio versus time data points for the standard briquette along
with the known thermal properties are used to calculate the surface co-
t efficient, h, in the following manner. A digital computer program is
’ written to compute Yc from Equations (6) through (231. By iteration
. and assuming various values of (h), the computed values of Yc can be
! made to converge on each of selected experimental Yc versus e data
i points. Thus, for a selected data point, the best experimental (h) is
that which when used in Equations (8) and (19) results in equal values
for the computed and experimental Yc values.
\
t For low conductivity test materials, the same method is used to
determine the best experimental value of k by using the h determined for
\ the standard and the other properties of the test material. If the test
1 material is a good conductor as discussed in the theory section, then
t
experience has shown that h should be computed from the experimental
cooling curve and then this value is used tommpute k by the same method
as for low conductivity test materials.

Discussion and Results
560
Three briquettes of aluminum, lead and silver were made to
test the validity of the experimental technique since their thermal
properties were available from the literature as shown in Table I.
Sintered, dense hematite (Fe 0 ) and a briquette of porous carbon
2.3 I
made from a partially devolatized coal were used as test materials
For these materials, all properties except the thermal Conductivities
shown in Table I were previously measured. Cooling curves for each
briquette were measured for a nitrogen flow rate of 0.9 scfm. Sur-
face heat transfer coefficients for lead, silver and alumixC
- vere
calculated by the method discussed in the data processing section.
For these materials , the llterature conductivity values were used to
calculate the surface coefficient. Table I shows that the calculated
or experiment@ h values for each metal are nearly identical.
This result is consistent with the theoretical basis of the experiment
and may be considered as establishing the validity of the method.
Also as additional evidence, aluminum was choosen as the standard and
k values for lead and silver were calculated using the h value for aluminum.
Table I shows that the calculated or experimental k values were
within 0.5 percent of the literature values. The conductivities for
hematite and porous carbon were calculated using aluminum as the standard.
Figure 3 shows the experimental data points with the solid lines ,
calculated from the theory. Note that the line for the carbon is curved
whereas those for the metals and hematite are linear. As discuss- '
ed previously, a linear cooling curve is obtained if the surface to
solid thermal resistance ratios are relatively large.
Note that for the
metals, lead which has the lowest conductivity and thermal diffusivity
cooled the fastest. This result is explained by examination of eqn.
(25) which shows that for similar gas flows and briquette dimensions,
the rate of cooling for different materials is determined by the heat
content, pC . It can be seen in Table I that the heat content for lead
is the lowegt of all metals tested.
. Summary
A rapid, simple method for determining thermal conductivity for a solid
has been developed. The solid can be either porous or non-porous and of
either high or low conductivity. If high conductivity materials are
tested, then both conductivity and heat capacity can be simultaneously
measured from one cooling experiment. The method was validated by using
the known thermal properties of lead, aluminum, and silver and the
experimental cooling curves in a comparsion with the computed results.
References
1- Newman, A. B., Industrial and Engineering Chemistry, Vol. 28, 1936, i
pp. 545-548
. .
_. .-. - . . . .
. .,
. .
I
1
i
1
I
1
1

1
I
A
An
''P
'H
I
'h
I
k
ma
"r
Q
R
r
' S
t
/X
tC
tS
xo
I xr
! yx
yr
( a
I
1 8
I
!
J
\
*I
e
-
551
Nomenclature
Half height of briquette; ft
Area; ft2
Coefficient in infinite series solution for temperature dis cri-
bution in Briquette
Specific heat; BTU/lb OF
Overall heat transfer coefficient; BTU/hr ft2 OF
Surface heat transfer coefficient; BTU/hr ft2 .OF
Thermal conductivity; BTU/hr ft2 O F / f t
Axial surface resistance; dimensionless
Radial surface resistance; dimensionless
Heat flux; BTU/hr
Maximum radius of briquette; ft
Radius of briquette; ft
Half width of infinite plate; ft
Temperature; OF
Temperature at center of briquette; OF
Initial temperature of briquette; OF
Temperature of cooling gas; OF
Distance of direction; ft
- - a8 Dimensionless time parameter for axial component
02
= a8 Dimensionless time parameter for radial component
r2
= Symbol for temperature ratio, axial component; dimensionless
= Symbol for temperature ratio, radial component; dimensionless
=(k/eCp) Thermal diffusivity; ft2/hr
= Time; minutes or hours
= Density; wft3

__- ._
562
Table I
THERMAL AND PHYSICAL PROPERTIES
Aluminum Silver Lead
a .01842 .02059 .01842
R
P
cP
PCP
h (experimental)
k (experimental)
k (literature)
a (experimental)
.04208 .04210
168.50 655.20
.2273 -0578
38.30 39.31
5.58 5.70
240.3
121.7 240.0
3.178* 6.113
* Average of literature sources
.04ii7
707.43
.0306
21.65
5.60
18.99
19.00
.8770
Hematite
.01842
.04208
306 .OO
-2090
63.95
5.58
12.10
none
.1892
.01958
.Of121 1
75.0
.2360 I
17.70
5.58
.0307
none
-00173
/
I

X
563
STANDARD CYLINDRICAL BRIQUETTE
Figure 1

ln
4
I
0
4J
v
\
1.0
0.8
0.6
L
-
0.4 -
0.3 -
0.2 -
-
0.1
-
.08 -
-
f
L M
:ooling
Tube LL
Legend
564
1 Thermocouple
I jl-. -2
t
rique tte
hl
EXPERIMENTAL APPARATUS
0- Lead
0- Silver
0- Aluminum
A- Porous Carbon Briquette '
* - Sintered hematite
Zlectri c
Furnace
I I I I I I I I I I
I
0 1
2 3 4 5
Time, 0, minutes
, Theoretical Cooling Curves and Experimental
Points for five Materials
T - L
c:::
* I
.
stop
Clock
Figure 2
Figuze 3